An emotion real-time monitoring method and system based on a smart bracelet
By collecting behavioral perception data in smart bracelets to generate a set of behavioral state features (Mov), the user's movement state is determined and the emotion analysis process is controlled. This solves the problem of misjudging fluctuations in physiological parameters during exercise as emotions, and improves the accuracy and reliability of emotion monitoring.
Patent Information
- Application Number
- CN202610307453.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-03-13
- Publication Date
- 2026-07-10
AI Technical Summary
Existing emotion monitoring methods based on smart bracelets are prone to misinterpreting fluctuations in physiological parameters caused by exercise as emotional states when users are exercising, leading to a decrease in the accuracy and reliability of emotion recognition results.
By collecting behavioral perception data generated during the user's wearing of the smart bracelet, a behavioral state feature set Mov is generated. Based on this, it is determined whether the user is in motion, and the triggering of the emotion analysis process is controlled. Emotion analysis is only performed when the user is not in motion, and the emotion analysis dataset Emo is collected and generated.
It effectively avoids the misjudgment of fluctuations in physiological parameters caused by exercise as emotional changes, improves the effectiveness and credibility of emotion analysis, and the output emotion monitoring results more realistically reflect the user's actual emotional state.
Smart Images

Figure CN122350706A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of wristband monitoring technology, specifically to a method and system for real-time emotion monitoring based on a smart wristband. Background Technology
[0002] With the widespread application of wearable devices in daily life, real-time emotion monitoring methods based on smart bracelets have gradually become an important research direction in the fields of health management and psychological state assessment. Smart bracelets can continuously collect various physiological and behavioral data such as heart rate and acceleration during daily wear, making it possible to monitor an individual's emotional state over a long period of time in natural life scenarios.
[0003] Existing emotion monitoring methods based on smart bracelets typically analyze physiological parameters such as heart rate and heart rate variability directly after data collection, and infer the user's emotional state based on parameter changes. However, in actual use, users are frequently engaged in activities such as walking, exercising, or physical labor, and these activities themselves cause significant changes in physiological parameters such as heart rate and respiratory rhythm. If emotion analysis is performed directly on these physiological parameters without effectively identifying and distinguishing the user's current activity state, it is highly likely that normal physiological fluctuations caused by exercise will be misjudged as emotional states such as tension, anxiety, or stress, leading to a significant decrease in the accuracy and reliability of emotion recognition results. Summary of the Invention
[0004] To address the shortcomings of existing technologies, this invention provides a method and system for real-time emotion monitoring based on a smart bracelet, which solves the problems mentioned in the background section.
[0005] To achieve the above objectives, the present invention provides a method for real-time emotion monitoring based on a smart bracelet, comprising the following steps: S1. Collect behavioral perception data generated during the user's wearing of the smart bracelet, and generate a behavioral state feature set Mov representing the user's physical activity state based on the behavioral perception data; S2. Based on the behavioral state feature set Mov, determine the user's current physical activity state and generate a motion determination result Sta of whether the user is in motion. S3. Based on the motion determination result Sta, execute the emotion determination trigger control to generate an emotion trigger result Trg indicating whether the emotion analysis process is allowed. S4. If the emotion triggering result Trg indicates that the emotion analysis process is allowed, collect physiological perception data and generate an emotion analysis dataset Emo for emotion analysis. S5. Perform sentiment analysis processing based on the sentiment analysis dataset Emo, and output the sentiment monitoring result Res representing the user's current emotional state.
[0006] Preferably, S1 includes S11; S11. During the process of the user wearing the smart bracelet, the accelerometer built into the smart bracelet continuously samples the movement state of the user's wrist in three-dimensional space to obtain raw behavioral perception data reflecting changes in the user's body activity; While collecting the raw behavioral perception data, the corresponding collection time information is recorded for each piece of raw behavioral perception data, and then time alignment processing is performed on the raw behavioral perception data with recorded collection time information. Using a preset time window as the time division benchmark, the raw behavioral perception data is mapped to the corresponding time window according to the collection time information, so that the raw behavioral perception data in the same time window corresponds to the same time reference; After completing the time alignment processing, the behavioral perception data in each time window are arranged and combined in chronological order to generate a behavioral perception data sequence Dat for subsequent behavioral state analysis.
[0007] Preferably, S1 further includes S12; S12. Based on the behavior perception data sequence Dat, perform activity feature extraction processing on the behavior perception data corresponding to each time window in sequence, so as to extract multi-dimensional activity features reflecting the user's physical activity state from the behavior perception data sequence Dat. The multidimensional features include: Within a single time window, the magnitude of change in behavioral perception data within the time window is analyzed, the degree of change in behavioral perception data within the time window is calculated, and the degree of change is used as the activity magnitude feature corresponding to the time window. Within the same time window, the changes in behavioral perception data within the time window are analyzed to determine the number of times the behavioral perception data changes within that time window, and the number of times is used as the activity frequency feature corresponding to the time window. Based on the changing trends of behavioral perception data corresponding to adjacent time windows, a continuity analysis is performed on the activity status within adjacent time windows to determine whether the activity status between adjacent time windows remains continuous or is interrupted, and the results of the continuity analysis are used as the activity continuity features corresponding to the adjacent time windows. The activity amplitude features, activity frequency features, and activity continuity features obtained within the same time window are combined to form a feature unit representing the physical activity state of that time window. Then, the feature units corresponding to each time window are aggregated in chronological order to finally generate a behavioral state feature set Mov representing the user's physical activity state.
[0008] Preferably, S2 includes S21; S21. Based on the behavioral state feature set Mov, select feature units corresponding to multiple consecutive time windows in the behavioral state feature set Mov as the judgment objects. The determination of the object is specifically completed by executing steps S211, S212 and S213; S211. Activity Intensity Determination: For each time window, read the activity amplitude characteristics within that time window and compare the activity amplitude characteristics with a preset amplitude determination benchmark to determine whether there is significant physical activity intensity within that time window; when the activity amplitude characteristics exceed the amplitude determination benchmark, determine that the time window is an effective intensity time window; when the activity amplitude characteristics do not exceed the amplitude determination benchmark, determine that the time window is an ineffective intensity time window. S212. Activity Frequency Determination: For the same time window, read the activity frequency characteristics within the time window and compare the activity frequency characteristics with a preset frequency determination benchmark to determine the frequency of physical activity within the time window; when the activity frequency characteristics exceed the frequency determination benchmark, determine the time window as a valid frequency time window; when the activity frequency characteristics do not exceed the frequency determination benchmark, determine the time window as an invalid frequency time window. S213. Activity Continuity Determination: Based on the activity continuity characteristics corresponding to adjacent time windows, determine whether the time windows that are simultaneously determined as intensity effective time windows and frequency effective time windows are continuously distributed in the time dimension among the multiple consecutive time windows; when the time windows are continuously distributed in the time dimension, it is determined that the physical activity is continuous; when the time windows are discontinuously distributed in the time dimension, it is determined that the physical activity is intermittent.
[0009] Preferably, after the judgment of the object is completed, the activity state type is determined: when multiple consecutive time windows are simultaneously determined to be effective time windows of intensity and frequency, and the activity continuity feature indicates that the physical activity is continuous, the physical activity state in the current time period is determined to be a continuous activity state; when a time window is determined to be effective time windows of intensity and frequency, and the activity continuity feature indicates that the physical activity is intermittent, the physical activity state in the current time period is determined to be an intermittent activity state; when multiple consecutive time windows are not determined to be effective time windows of intensity or frequency, the physical activity state in the current time period is determined to be a stable activity state. The determined physical activity state is output as the activity state analysis result Act, which represents the physical activity state within the current time period.
[0010] Preferably, S2 further includes S22; S22. Read the body activity state type from the activity state analysis result Act; when the activity state analysis result Act reflects a continuous or intermittent activity state within the current time period, determine that the user has physical movement behavior within the current time period and classify the current state as a movement state; when the activity state analysis result Act reflects a stable activity state within the current time period, determine that the user does not have physical movement behavior within the current time period and classify the current state as a non-movement state; and output the state determination result as a movement determination result Sta indicating whether the user is currently in a movement state.
[0011] Preferably, S3 includes S31; S31. Read the motion determination result Sta to determine whether the user is in motion during the current time period; When the motion determination result Sta indicates that the user is in motion, the emotion analysis process is prevented from being triggered, and an emotion triggering result Trg indicating that the emotion analysis process is prohibited from being executed is generated. When the motion determination result Sta indicates that the user is not in motion, the control emotion analysis process is allowed to be triggered, and an emotion trigger result Trg indicating that the emotion analysis process is allowed to be executed is generated.
[0012] Preferably, S4 includes S41; S41. Read the emotion triggering result Trg to determine whether to allow the emotion analysis process to be executed; When the emotion trigger result Trg indicates that the emotion analysis process is allowed, the physiological perception data collection process is initiated, and physiological perception data reflecting changes in the user's physiological state are collected through the physiological sensors built into the smart bracelet. During the collection of the physiological sensory data, the corresponding collection time information is recorded for each physiological sensory data, and the physiological sensory data is organized according to a preset time window. After completing the collection and time processing of the physiological perception data, the physiological perception data is collected in chronological order to generate the emotion analysis dataset Emo for emotion analysis processing. When the emotion trigger result Trg indicates that the emotion analysis process is not allowed, the process of collecting physiological perception data is stopped or skipped, and the emotion analysis dataset Emo is not generated.
[0013] Preferably, S5 includes S51; S51. Based on the emotion analysis dataset Emo, read the physiological perception data collected during the time period during which the emotion analysis process is allowed to be executed; Emotion feature extraction processing is performed on the physiological perception data in the emotion analysis dataset Emo to obtain emotion-related features that reflect changes in the user's physiological state; Based on the aforementioned emotion-related features, the user's current emotional state is determined to obtain a determination result representing the user's current emotional state. The determination result is then output as an emotion monitoring result Res, representing the user's current emotional state.
[0014] A real-time emotion monitoring system based on a smart bracelet includes a behavior perception module, an activity state determination module, an emotion determination trigger module, an analysis process execution module, and an emotion processing module. The behavior perception module collects behavior perception data generated during the user's wearing of the smart bracelet, and generates a set of behavior state features (Mov) representing the user's physical activity state based on the behavior perception data. The activity state determination module determines the user's current physical activity state based on the behavior state feature set Mov, and generates a motion determination result Sta indicating whether the user is in motion. The emotion determination triggering module executes emotion determination triggering control based on the motion determination result Sta, and generates an emotion triggering result Trg indicating whether the emotion analysis process is allowed. When the emotion triggering result Trg indicates that the emotion analysis process is allowed, the analysis process execution module collects physiological perception data and generates an emotion analysis dataset Emo for emotion analysis. The emotion processing module performs emotion analysis processing based on the emotion analysis dataset Emo, and outputs the emotion monitoring result Res, which represents the user's current emotional state.
[0015] This invention provides a method and system for real-time emotion monitoring based on a smart bracelet, which has the following beneficial effects: (1) By collecting behavioral perception data generated during the user's wearing of the smart bracelet, a behavioral state feature set Mov representing the user's physical activity state is constructed, and a motion judgment result Sta is generated on this basis, so as to clearly distinguish whether the user is in a motion state before the emotion analysis process is started. Subsequently, an emotion triggering result Trg is generated based on the motion judgment result Sta, so that the emotion analysis process is only triggered when the user is not in a motion state, thus avoiding the problem of fluctuations in physiological parameters such as heart rate caused by walking, running or physical activity being misjudged as emotional changes from the process level; Physiological perception data is collected only within the time period indicated by the emotion trigger result Trg, allowing the emotion analysis process to be executed, and an emotion analysis dataset Emo is generated. This ensures that the data in the emotion analysis dataset Emo originates from physiological state ranges that better align with the preconditions for emotion analysis, thereby improving the effectiveness and reliability of subsequent emotion analysis. Ultimately, the emotion monitoring result Res output based on the emotion analysis dataset Emo can more realistically reflect the user's actual emotional state in daily life.
[0016] (2) By introducing an emotion determination triggering control mechanism with the motion determination result Sta as a prerequisite in the real-time emotion monitoring process, the emotion analysis process is no longer a continuously running default process, but a controlled process that is only allowed to be executed under specific physiological conditions. Specifically, this solution reads the motion determination result Sta and generates an emotion triggering result Trg indicating that the emotion analysis process is prohibited when the user is in motion, directly blocking subsequent emotion analysis based on physiological perception data at the process level; while when the user is not in motion, an emotion triggering result Trg indicating that the emotion analysis process is allowed to be executed is generated, so that emotion analysis is only triggered during the time period when physiological signals are more likely to reflect the true emotional state.
[0017] (3) By using the emotion trigger result Trg as the sole prerequisite for collecting physiological perception data, physiological perception data is collected only within the time period allowed for executing the emotion analysis process, and an emotion analysis dataset Emo is generated accordingly. This ensures from the source that the data in the emotion analysis dataset Emo originates from a physiological state range where the user is not in motion and which better meets the prerequisites for emotion analysis. Simultaneously, by recording the collection time information for the physiological perception data and organizing it according to a preset time window, the emotion analysis dataset Emo maintains consistency with the preceding behavior judgment steps in terms of time structure, providing a clear temporal correspondence for subsequent analysis. Attached Figure Description
[0018] Figure 1 This is a schematic diagram illustrating the steps of a real-time emotion monitoring method based on a smart bracelet according to the present invention. Figure 2 This is a schematic diagram of a real-time emotion monitoring system based on a smart bracelet according to the present invention. Detailed Implementation
[0019] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention. Example 1
[0020] This invention provides a method for real-time emotion monitoring based on a smart bracelet. Please refer to [link / reference]. Figure 1 This includes the following steps: S1. Collect behavioral perception data generated during the user's wearing of the smart bracelet, and generate a behavioral state feature set Mov representing the user's physical activity state based on the behavioral perception data; S2. Based on the behavioral state feature set Mov, determine the user's current physical activity state and generate a motion determination result Sta of whether the user is in motion. S3. Based on the motion determination result Sta, execute the emotion determination trigger control to generate an emotion trigger result Trg indicating whether the emotion analysis process is allowed. S4. If the emotion triggering result Trg indicates that the emotion analysis process is allowed, collect physiological perception data and generate an emotion analysis dataset Emo for emotion analysis. S5. Perform sentiment analysis processing based on the sentiment analysis dataset Emo, and output the sentiment monitoring result Res representing the user's current emotional state.
[0021] In this embodiment, a real-time emotion monitoring method based on a smart bracelet, constructed through the aforementioned steps S1 to S5, achieves reliable, controlled, and feasible real-time monitoring of the user's emotional state without increasing the hardware structure of the smart bracelet. Specifically, this solution first collects behavioral perception data generated during the user's wearing of the smart bracelet, constructs a behavioral state feature set Mov representing the user's physical activity state, and generates a motion determination result Sta based on this, thus clearly distinguishing whether the user is in a state of motion before the emotion analysis process is initiated. Subsequently, an emotion triggering result Trg is generated based on the motion determination result Sta, ensuring that the emotion analysis process is triggered only when the user is not in a state of motion, thus avoiding the problem of fluctuations in physiological parameters such as heart rate caused by walking, running, or physical activity being misjudged as emotional changes at the process level.
[0022] Based on this, physiological perception data is collected only within the time period indicated by the emotion triggering result Trg, during which the emotion analysis process is allowed, and an emotion analysis dataset Emo is generated. This ensures that the data in the emotion analysis dataset Emo originates from physiological state ranges that better match the preconditions for emotion analysis, thereby improving the effectiveness and reliability of subsequent emotion analysis. Ultimately, the emotion monitoring result Res output based on the emotion analysis dataset Emo can more realistically reflect the user's actual emotional state in daily life. For example, during a user's commute, walking, or climbing stairs, the emotion analysis process will not be triggered because the motion judgment result Sta indicates that the user is in motion, thus avoiding the misidentification of increased heart rate caused by exercise as tension or anxiety. The emotion analysis process is only allowed to execute in non-motion scenarios such as sitting at work, resting, or relaxing before bed, making the output emotion monitoring result Res more consistent with the user's true emotional experience. Therefore, this solution effectively solves the problems of emotion monitoring results being easily affected by motion interference, high misjudgment rate, and insufficient long-term monitoring stability in existing technologies, significantly improving the reliability and practical value of real-time emotion monitoring based on smart bracelets in real-world application scenarios. Example 2
[0023] Specifically: S1 includes S11; S11. During the process of the user wearing the smart bracelet, the accelerometer built into the smart bracelet continuously samples the movement state of the user's wrist in three-dimensional space to obtain raw behavioral perception data reflecting changes in the user's body activity; While collecting the raw behavioral perception data, the corresponding collection time information is recorded for each piece of raw behavioral perception data, and then time alignment processing is performed on the raw behavioral perception data with recorded collection time information. Using a preset time window as the time division benchmark, the raw behavioral perception data is mapped to the corresponding time window according to the collection time information, so that the raw behavioral perception data in the same time window corresponds to the same time reference; After completing the time alignment processing, the behavioral perception data in each time window are arranged and combined in chronological order to generate a behavioral perception data sequence Dat for subsequent behavioral state analysis; It should be noted that: The time alignment process uses the system time of the smart bracelet as a unified time reference and a preset time window as the time division unit. For each piece of raw behavior perception data with recorded collection time information, the raw behavior perception data is mapped to the corresponding time window according to the collection time information. When there are multiple pieces of raw behavior perception data in the same time window, the multiple pieces of raw behavior perception data are grouped into the same time window to form a time window data unit. When there is no raw behavior perception data in a certain time window, the time window is kept empty and the corresponding status information is recorded, or the valid raw behavior perception data in the adjacent time window is used for completion processing.
[0024] S1 further includes S12; S12. Based on the behavior perception data sequence Dat, perform activity feature extraction processing on the behavior perception data corresponding to each time window in sequence, so as to extract multi-dimensional activity features reflecting the user's physical activity state from the behavior perception data sequence Dat. The multidimensional features include: Within a single time window, the magnitude of change in behavioral perception data within the time window is analyzed, the degree of change in behavioral perception data within the time window is calculated, and the degree of change is used as the activity magnitude feature corresponding to the time window. Within the same time window, the changes in behavioral perception data within the time window are analyzed to determine the number of times the behavioral perception data changes within that time window, and the number of times is used as the activity frequency feature corresponding to the time window. Based on the changing trends of behavioral perception data corresponding to adjacent time windows, a continuity analysis is performed on the activity status within adjacent time windows to determine whether the activity status between adjacent time windows remains continuous or is interrupted, and the results of the continuity analysis are used as the activity continuity features corresponding to the adjacent time windows. The activity amplitude features, activity frequency features, and activity continuity features obtained within the same time window are combined to form a feature unit representing the physical activity state of that time window; then the feature units corresponding to each time window are aggregated in chronological order to finally generate a behavioral state feature set Mov representing the user's physical activity state. It should be noted that: The activity amplitude feature, activity frequency feature, and activity continuity feature are all structured activity features extracted from the behavior perception data sequence Dat under time window constraints, which are used to reflect the user's physical activity status from different dimensions. The activity amplitude feature refers to: within a single time window, performing an overall change analysis on the behavioral perception data within that time window, and calculating at least one of the maximum change, average change, or cumulative change of the behavioral perception data within that time window to obtain a numerical result that reflects the intensity of physical activity within that time window, and defining this numerical result as the activity amplitude feature corresponding to the time window; thereby enabling the activity amplitude feature to stably reflect the intensity level of physical activity per unit time, without being affected by a single abnormal sampling point; The activity frequency feature refers to: within a single time window, statistically analyzing the changes in behavioral perception data within that time window, identifying the number of effective changes in the behavioral perception data within that time window based on preset change judgment rules, and defining the number of changes or the frequency value calculated from the number of changes as the activity frequency feature corresponding to that time window; thereby enabling the activity frequency feature to reflect the frequency of a user's physical activities within a unit of time. The activity continuity feature refers to: based on the corresponding behavioral perception data or extracted activity features within adjacent time windows, comparing and analyzing the trend of activity state changes between adjacent time windows to determine whether the body activity state remains continuous in the time dimension; when the activity state within adjacent time windows remains consistent or the change range is within a continuous range, the activity state is determined to be continuous; when the activity state within adjacent time windows experiences a significant interruption or abrupt change, the activity state is determined to be discontinuous; and the continuity judgment result or the continuity degree value calculated from the continuity judgment result is defined as the activity continuity feature corresponding to the adjacent time windows; The activity amplitude features, activity frequency features, and activity continuity features extracted through the above methods provide a complete characterization of the user's physical activity state from three dimensions: activity intensity, activity frequency, and activity duration. These three types of features are then combined and aggregated according to time windows to form a behavioral state feature set Mov that is structurally clear, time-traceable, and actionable. This provides a stable and reliable feature foundation for generating motion determination results Sta based on the behavioral state feature set Mov.
[0025] In this embodiment, through the coordinated execution of steps S11 and S12, this solution constructs a user body activity representation mechanism with a clear structure, consistent timing, and multi-dimensional characterization capabilities at the foundational level of behavior perception data. Specifically, by performing time alignment processing on the original behavior perception data and generating a behavior perception data sequence Dat, the behavior perception data from continuous sampling has a unified reference in the time dimension, thereby avoiding data jumps or misinterpretations of behavior states caused by inconsistent sampling timing. Based on this, activity amplitude features, activity frequency features, and activity continuity features are further extracted from the behavior perception data sequence Dat, and the three types of features are combined according to time windows to form a behavior state feature set Mov, enabling the user's body activity state to be fully characterized from three complementary dimensions: activity intensity, activity frequency, and activity duration.
[0026] Compared to existing methods that rely solely on a single activity level indicator or simple threshold judgment, this approach, through the aforementioned behavioral state feature set (Mov), can more precisely distinguish various common but easily confused behavioral states in real life. For example, in scenarios where a user works at a desk for extended periods but occasionally adjusts their posture, the activity amplitude characteristics may show brief changes. However, because the activity frequency and activity continuity characteristics do not form a sustained response, the generated behavioral state feature set (Mov) can accurately reflect that this state still belongs to low-activity behavior. In scenarios where a user goes up and down stairs or performs housework, the three types of features form a consistent response across multiple time windows, resulting in a stable characterization of the body's activity state. Therefore, by constructing a behavioral state feature set (Mov) with time traceability and multi-dimensional judgment criteria at the source stage, this approach provides a more stable, interpretable, and realistic input foundation for the subsequent generation of the movement judgment result (Sta), significantly improving the adaptability and engineering usability of the entire real-time emotion monitoring method in complex daily environments. Example 3
[0027] Specifically: S2 includes S21; S21. Based on the behavioral state feature set Mov, select feature units corresponding to multiple consecutive time windows in the behavioral state feature set Mov as the judgment objects. The determination of the object is specifically completed by executing steps S211, S212 and S213; S211. Activity Intensity Determination: For each time window, read the activity amplitude characteristics within that time window and compare the activity amplitude characteristics with a preset amplitude determination benchmark to determine whether there is significant physical activity intensity within that time window; when the activity amplitude characteristics exceed the amplitude determination benchmark, determine that the time window is an effective intensity time window; when the activity amplitude characteristics do not exceed the amplitude determination benchmark, determine that the time window is an ineffective intensity time window. S212. Activity Frequency Determination: For the same time window, read the activity frequency characteristics within the time window and compare the activity frequency characteristics with a preset frequency determination benchmark to determine the frequency of physical activity within the time window; when the activity frequency characteristics exceed the frequency determination benchmark, determine the time window as a valid frequency time window; when the activity frequency characteristics do not exceed the frequency determination benchmark, determine the time window as an invalid frequency time window. S213. Activity Continuity Determination: Based on the activity continuity characteristics corresponding to adjacent time windows, determine whether the time windows that are simultaneously determined as intensity effective time windows and frequency effective time windows are continuously distributed in the time dimension among the multiple consecutive time windows; when the time windows are continuously distributed in the time dimension, it is determined that the physical activity is continuous; when the time windows are discontinuously distributed in the time dimension, it is determined that the physical activity is intermittent.
[0028] After the judgment of the object is completed, the activity state type is determined: when multiple consecutive time windows are simultaneously judged as effective time windows of intensity and frequency, and the activity continuity feature indicates that the physical activity is continuous, the physical activity state in the current time period is determined to be a continuous activity state; when a time window is judged as an effective time window of intensity and frequency, and the activity continuity feature indicates that the physical activity is intermittent, the physical activity state in the current time period is determined to be an intermittent activity state; when multiple consecutive time windows are not judged as effective time windows of intensity or frequency, the physical activity state in the current time period is determined to be a stable activity state. The determined physical activity state is output as the activity state analysis result Act, which represents the physical activity state within the current time period. It should be noted that: The activity amplitude feature is used to reflect the intensity of body activity within a single time window. By comparing the activity amplitude feature with the amplitude judgment benchmark, it is possible to effectively distinguish between minute movements in a static state and real body activity behavior, thereby avoiding misjudgment caused by slight shaking or noise. The activity frequency feature is used to reflect the number of times physical activity occurs within a single time window. By comparing the activity frequency feature with the frequency determination benchmark, it is possible to distinguish between occasional actions and continuous physical activity behaviors, thereby avoiding misidentifying occasional actions as a state of motion. The activity continuity feature is used to reflect the continuity of physical activity in the time dimension. By analyzing the distribution of time windows that are determined to be effective in intensity and frequency on the time axis, it can be determined whether physical activity continues to exist in multiple adjacent time windows, thereby distinguishing between continuous activity state and intermittent activity state.
[0029] S2 further includes S22; S22. Read the body activity state type from the activity state analysis result Act; when the activity state analysis result Act reflects a continuous or intermittent activity state within the current time period, determine that the user has physical movement behavior within the current time period and classify the current state as a movement state; when the activity state analysis result Act reflects a stable activity state within the current time period, determine that the user does not have physical movement behavior within the current time period and classify the current state as a non-movement state; and output the state determination result as a movement determination result Sta indicating whether the user is currently in a movement state.
[0030] In this embodiment, through the execution of steps S21 and S22, this solution establishes a motion state recognition mechanism based on multi-feature collaboration and temporal continuity constraints at the behavioral state determination level. Specifically, this solution does not directly determine whether a user is in a motion state based on a single indicator. Instead, it first determines the activity amplitude feature, activity frequency feature, and activity continuity feature step by step based on the behavioral state feature set Mov, and then forms a clear activity state analysis result Act based on this, and further generates a motion determination result Sta. By using activity intensity, the number of times the activity occurs, and the continuity of the activity in the time dimension as mutually constraining determination conditions, the recognition of motion state no longer depends on instantaneous fluctuations, but is based on consistent performance across multiple time windows.
[0031] This determination method has significant advantages in real-world wearing scenarios. For example, when a user briefly raises their hand to pick up an object or occasionally shakes their wrist, although the amplitude of the activity may briefly exceed the amplitude determination benchmark, the activity frequency and continuity features fail to simultaneously meet the determination conditions within multiple consecutive time windows. Therefore, the generated activity state analysis result Act will still reflect a stable activity state, allowing the output motion determination result Sta to accurately indicate a non-motion state. In real-world motion scenarios such as brisk walking, climbing stairs, or performing continuous housework, the activity amplitude and frequency features remain valid across multiple adjacent time windows, and the activity continuity feature indicates that the activity state exists continuously over time. This ensures that the generated motion determination result Sta stably reflects that the user is indeed in a motion state. Thus, by introducing multi-dimensional feature collaboration and temporal continuity constraints in the determination stage, this solution effectively reduces the probability of misjudging the motion state caused by short-term actions, sporadic behaviors, or noise interference. It significantly improves the accuracy and engineering reliability of motion state determination in complex daily environments, providing a more robust basis for subsequent emotion-triggered control based on the motion determination result Sta. Example 4
[0032] Specifically: S3 includes S31; S31. Read the motion determination result Sta to determine whether the user is in motion during the current time period; When the motion determination result Sta indicates that the user is in motion, the emotion analysis process is prevented from being triggered, and an emotion triggering result Trg indicating that the emotion analysis process is prohibited from being executed is generated. When the motion determination result Sta indicates that the user is not in motion, the control emotion analysis process is allowed to be triggered, and an emotion trigger result Trg indicating that the emotion analysis process is allowed to be executed is generated; The emotion triggering result Trg is used to control whether the subsequent emotion analysis process based on physiological perception data is executed. It should be noted that: When the motion determination result Sta indicates that the user is in motion, the user's physiological parameters such as heart rate are mainly affected by physical movement. If the emotion analysis process is executed at this time, it is easy to misjudge the physiological changes caused by physical movement as emotional fluctuations. Therefore, when the motion determination result Sta indicates that the user is in motion, an emotion trigger result Trg is generated to indicate that the emotion analysis process is prohibited from being executed, thereby blocking the execution of the subsequent emotion analysis process and avoiding the generation of invalid or erroneous emotion determination results. When the motion determination result Sta indicates that the user is not in motion, the changes in the user's physiological parameters are more likely to be related to the emotional state. Therefore, when the motion determination result Sta indicates that the user is not in motion, by generating an emotion trigger result Trg that indicates that the emotion analysis process is allowed to be executed, the subsequent emotion analysis process based on physiological perception data is only executed when there are valid preconditions. By using the above-mentioned emotion determination trigger control method based on motion determination result Sta, the execution of the emotion analysis process has clear preconditions, thereby improving the reliability and accuracy of emotion monitoring results in practical application scenarios without increasing hardware.
[0033] In this embodiment, through the execution of step S31, this solution introduces an emotion determination triggering control mechanism in the real-time emotion monitoring process, which uses the motion determination result Sta as a prerequisite. This prevents the emotion analysis process from being a continuously running default process, but rather a controlled process that is only allowed to be executed under specific physiological conditions. Specifically, this solution reads the motion determination result Sta and generates an emotion triggering result Trg indicating that the emotion analysis process is prohibited when the user is in motion, directly blocking subsequent emotion analysis based on physiological perception data at the process level. When the user is not in motion, an emotion triggering result Trg indicating that the emotion analysis process is allowed is generated, ensuring that emotion analysis is only triggered during the time period when physiological signals are more likely to reflect the true emotional state.
[0034] This control method offers significant advantages in practical applications. For example, during daily commutes, brisk walking, climbing stairs, or doing housework, even if physiological parameters such as heart rate collected by the smart bracelet fluctuate significantly, the generated motion determination result Sta indicates that the user is in motion, preventing the emotion analysis process from being triggered. This avoids misidentifying physiological changes caused by physical movement as emotional states such as tension or anxiety. Conversely, in scenarios where the user is resting after exercise, focusing on work, or relaxing before bed, the motion determination result Sta switches to a non-motion state, allowing the emotion analysis process to execute via the emotion trigger result Trg. This makes the final emotion-related results more closely reflect the user's true emotional experience. Therefore, by introducing an emotion determination trigger mechanism based on the motion determination result Sta at the process control level, this solution effectively reduces the number of invalid emotion analysis executions, lowers the probability of misjudgment, and significantly improves the reliability, stability, and practical application value of emotion monitoring results in complex daily use environments without increasing hardware costs. Example 5
[0035] Specifically: S4 includes S41; S41. Read the emotion triggering result Trg to determine whether to allow the emotion analysis process to be executed; When the emotion trigger result Trg indicates that the emotion analysis process is allowed, the physiological perception data collection process is initiated, and physiological perception data reflecting changes in the user's physiological state are collected through the physiological sensors built into the smart bracelet. During the collection of the physiological sensing data, the corresponding collection time information is recorded for each physiological sensing data, and the physiological sensing data is organized according to a preset time window so that the physiological sensing data is consistent with the previous processing steps in the time dimension. After completing the collection and time processing of the physiological perception data, the physiological perception data is collected in chronological order to generate the emotion analysis dataset Emo for emotion analysis processing. When the emotion triggering result Trg indicates that the emotion analysis process is not allowed, the process of collecting physiological perception data is stopped or skipped, and the emotion analysis dataset Emo is not generated. It should be noted that: The emotion analysis dataset Emo refers to the set of physiological perception data collected and formed within the time period during which the emotion analysis process is allowed to be executed, under the control of the emotion triggering result Trg. The physiological perception data refers to data acquired by the built-in physiological sensors of the smart bracelet that can reflect changes in the user's physiological state, including but not limited to heart rate-related data. By collecting the physiological perception data only when the emotion trigger result Trg indicates that the emotion analysis process is allowed, the data in the emotion analysis dataset Emo comes from the time period when the user is not in a state of motion, thereby reducing the interference of physical movement on the physiological perception data. Meanwhile, by recording and collecting time information for the physiological perception data and organizing it according to a preset time window, the emotion analysis dataset Emo is kept consistent with the data used in the previous steps in terms of time structure, thereby ensuring that the subsequent emotion analysis processing based on the emotion analysis dataset Emo has a clear time correspondence and a stable data foundation. The Emo dataset, generated through the controlled collection and data organization methods described above, can provide physiological data input with effective prerequisites, temporal consistency, and feasibility for subsequent emotion monitoring without increasing the hardware of the smart bracelet.
[0036] S5 includes S51; S51. Based on the emotion analysis dataset Emo, read the physiological perception data collected during the time period during which the emotion analysis process is allowed to be executed; Emotion feature extraction processing is performed on the physiological perception data in the emotion analysis dataset Emo to obtain emotion-related features that reflect changes in the user's physiological state; Based on the aforementioned emotion-related features, the user's current emotional state is determined to obtain a determination result representing the user's current emotional state. The determination result is then output as an emotion monitoring result Res, representing the user's current emotional state. It should be noted that: The emotion monitoring result Res is an emotional state output result generated based on the emotion analysis dataset Emo, following the common data processing procedures in existing emotion analysis technologies. The emotion feature extraction process refers to: based on the physiological perception data contained in the emotion analysis dataset Emo, performing time series analysis on the physiological perception data to obtain feature data that reflects changes in physiological state; wherein, the emotion feature extraction process includes, but is not limited to, at least one of the following processing methods: performing statistical analysis on the change trend of the physiological perception data within a preset time window to obtain statistical features reflecting the level of physiological change; comparing and analyzing the changes of the physiological perception data between adjacent time windows to obtain change features reflecting the rate of physiological change; analyzing the fluctuation of the physiological perception data in the time dimension to obtain fluctuation features reflecting the stability of physiological state; through the above time series-based processing methods, the original physiological perception data is converted into structured emotion-related feature data; The emotion state determination process refers to: based on the emotion-related feature data obtained from the emotion feature extraction process, matching or mapping the emotion-related feature data according to preset emotion determination rules to determine the user's current emotion state; wherein, the emotion determination rules are a set of rules used in existing emotion analysis technologies, used to map different feature combinations to different emotion state types; when the emotion-related feature data satisfies the determination rule corresponding to a certain emotion state, the user's current emotion state is determined to be that emotion state, and the determination result is output as the emotion monitoring result Res; By employing the aforementioned emotion feature extraction and emotion state determination processes based on existing emotion analysis technologies, the generation of the emotion monitoring result Res is not dependent on a specific algorithm implementation. Instead, it is completed under the premise of having clear processing steps and judgment logic, thereby ensuring the feasibility of this step in practical applications and avoiding limiting the scope of protection of this solution to specific emotion analysis algorithms.
[0037] In this embodiment, through the coordinated execution of steps S41 and S51, this solution constructs a controlled, time-consistent, and highly universal emotion analysis output mechanism at the end of real-time emotion monitoring. Specifically, this solution uses the emotion trigger result Trg as the sole prerequisite for collecting physiological perception data, collecting physiological perception data only within the time period allowed to execute the emotion analysis process, and generating an emotion analysis dataset Emo accordingly. This ensures from the source that the data in the emotion analysis dataset Emo originates from a physiological state range where the user is not in motion and that better meets the prerequisites for emotion analysis. Simultaneously, by recording the collection time information for the physiological perception data and organizing it according to a preset time window, the emotion analysis dataset Emo maintains consistency with the preceding behavior judgment steps in terms of time structure, providing a clear temporal correspondence for subsequent analysis.
[0038] Based on this, this solution uses the aforementioned emotion analysis dataset Emo and employs common emotion feature extraction and emotion state determination processes found in existing emotion analysis technologies to output an emotion monitoring result Res representing the user's current emotional state. Since the generation of the emotion monitoring result Res does not depend on specific emotion analysis algorithms or fixed models, but is built upon standardized data structures and clear determination processes, this solution can flexibly adapt to various emotion analysis strategies in different application scenarios. For example, in non-exercise scenarios such as daily work, reading, or rest, the generated emotion analysis dataset Emo can continuously and stably reflect changes in the user's physiological state, making the output emotion monitoring result Res more suitable for long-term emotion trend analysis. In complex scenarios where users frequently switch activity states, the emotion analysis dataset Emo is not generated for time periods that do not meet the conditions, avoiding a large amount of meaningless or interfering data from entering the emotion analysis process. Therefore, without increasing the hardware cost of the smart bracelet, this solution achieves the purification of the emotion analysis data source, the consistency of the time structure, and the high universality of the emotion monitoring result output, significantly improving the engineering usability and long-term application value of real-time emotion monitoring based on smart bracelets in real-world usage environments. Example 6
[0039] A real-time emotion monitoring system based on a smart bracelet, please refer to... Figure 2 Specifically, it includes a behavior perception module, an activity status determination module, an emotion determination trigger module, an analysis process execution module, and an emotion processing module; The behavior perception module collects behavior perception data generated during the user's wearing of the smart bracelet, and generates a set of behavior state features (Mov) representing the user's physical activity state based on the behavior perception data. The activity state determination module determines the user's current physical activity state based on the behavior state feature set Mov, and generates a motion determination result Sta indicating whether the user is in motion. The emotion determination triggering module executes emotion determination triggering control based on the motion determination result Sta, and generates an emotion triggering result Trg indicating whether the emotion analysis process is allowed. When the emotion triggering result Trg indicates that the emotion analysis process is allowed, the analysis process execution module collects physiological perception data and generates an emotion analysis dataset Emo for emotion analysis. The emotion processing module performs emotion analysis processing based on the emotion analysis dataset Emo, and outputs the emotion monitoring result Res, which represents the user's current emotional state.
[0040] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended technical solutions and their equivalents.
Claims
1. A method for real-time emotion monitoring based on a smart bracelet, characterized in that: Includes the following steps: S1. Collect behavioral perception data generated during the user's wearing of the smart bracelet, and generate a behavioral state feature set Mov representing the user's physical activity state based on the behavioral perception data; S2. Based on the behavioral state feature set Mov, determine the user's current physical activity state and generate a motion determination result Sta of whether the user is in motion. S3. Based on the motion determination result Sta, execute the emotion determination trigger control to generate an emotion trigger result Trg indicating whether the emotion analysis process is allowed. S4. If the emotion triggering result Trg indicates that the emotion analysis process is allowed, collect physiological perception data and generate an emotion analysis dataset Emo for emotion analysis. S5. Perform sentiment analysis processing based on the sentiment analysis dataset Emo, and output the sentiment monitoring result Res representing the user's current emotional state.
2. The method for real-time emotion monitoring based on a smart bracelet according to claim 1, characterized in that: S1 includes S11; S11. During the process of the user wearing the smart bracelet, the accelerometer built into the smart bracelet continuously samples the movement state of the user's wrist in three-dimensional space to obtain raw behavioral perception data reflecting changes in the user's body activity; while collecting the raw behavioral perception data, the corresponding collection time information is recorded for each piece of raw behavioral perception data, and then time alignment processing is performed on the raw behavioral perception data with the recorded collection time information. Using a preset time window as the time division benchmark, the raw behavioral perception data is mapped to the corresponding time window according to the collection time information, so that the raw behavioral perception data located in the same time window corresponds to the same time reference. After completing the time alignment process, the behavior-aware data within each time window are arranged and combined in chronological order to generate a behavior-aware data sequence Dat for subsequent behavior state analysis.
3. The method for real-time emotion monitoring based on a smart bracelet according to claim 2, characterized in that: S1 further includes S12; S12. Based on the behavior perception data sequence Dat, perform activity feature extraction processing on the behavior perception data corresponding to each time window in sequence, so as to extract multi-dimensional activity features reflecting the user's physical activity state from the behavior perception data sequence Dat. The multidimensional features include: Within a single time window, the magnitude of change in behavioral perception data within the time window is analyzed, the degree of change in behavioral perception data within the time window is calculated, and the degree of change is used as the activity magnitude feature corresponding to the time window. Within the same time window, the changes in behavioral perception data within the time window are analyzed to determine the number of times the behavioral perception data changes within that time window, and the number of times is used as the activity frequency feature corresponding to the time window. Based on the changing trends of behavioral perception data corresponding to adjacent time windows, a continuity analysis is performed on the activity status within adjacent time windows to determine whether the activity status between adjacent time windows remains continuous or is interrupted, and the results of the continuity analysis are used as the activity continuity features corresponding to the adjacent time windows. The activity amplitude features, activity frequency features, and activity continuity features obtained within the same time window are combined to form a feature unit representing the physical activity state of that time window; then the feature units corresponding to each time window are aggregated in chronological order to finally generate a behavioral state feature set Mov representing the user's physical activity state.
4. The method for real-time emotion monitoring based on a smart bracelet according to claim 3, characterized in that: S2 includes S21; S21. Based on the behavioral state feature set Mov, select feature units corresponding to multiple consecutive time windows in the behavioral state feature set Mov as the judgment objects. The determination of the object is specifically completed by executing steps S211, S212 and S213; S211. Activity Intensity Determination: For each time window, read the activity amplitude characteristics within that time window and compare the activity amplitude characteristics with a preset amplitude determination benchmark to determine whether there is significant physical activity intensity within that time window; when the activity amplitude characteristics exceed the amplitude determination benchmark, determine that the time window is an effective intensity time window; when the activity amplitude characteristics do not exceed the amplitude determination benchmark, determine that the time window is an ineffective intensity time window. S212. Activity Frequency Determination: For the same time window, read the activity frequency characteristics within the time window and compare the activity frequency characteristics with a preset frequency determination benchmark to determine the frequency of physical activity within the time window; when the activity frequency characteristics exceed the frequency determination benchmark, determine the time window as a valid frequency time window; when the activity frequency characteristics do not exceed the frequency determination benchmark, determine the time window as an invalid frequency time window. S213. Activity continuity determination: Based on the activity continuity characteristics corresponding to adjacent time windows, determine whether the time windows that are simultaneously determined as intensity effective time windows and frequency effective time windows are continuously distributed in the time dimension among the multiple consecutive time windows; when the time windows are continuously distributed in the time dimension, it is determined that the physical activity is continuous; when the time windows are discontinuously distributed in the time dimension, it is determined that the physical activity is intermittent.
5. The method for real-time emotion monitoring based on a smart bracelet according to claim 4, characterized in that: After the judgment of the object is completed, the activity state type is determined: when multiple consecutive time windows are simultaneously judged as effective time windows of intensity and frequency, and the activity continuity feature indicates that the physical activity is continuous, the physical activity state in the current time period is determined to be a continuous activity state; when a time window is judged as an effective time window of intensity and frequency, and the activity continuity feature indicates that the physical activity is intermittent, the physical activity state in the current time period is determined to be an intermittent activity state; when multiple consecutive time windows are not judged as effective time windows of intensity or frequency, the physical activity state in the current time period is determined to be a stable activity state. The determined physical activity state is output as the activity state analysis result Act, which represents the physical activity state within the current time period.
6. The method for real-time emotion monitoring based on a smart bracelet according to claim 5, characterized in that: S2 further includes S22; S22. Read the body activity state type in the activity state analysis result Act; when the activity state analysis result Act reflects that the body activity state in the current time period is a continuous activity state or an intermittent activity state, determine that the user has body movement behavior in the current time period, and determine the current state as a movement state; when the activity state analysis result Act reflects that the body activity state in the current time period is a stable activity state, determine that the user does not have body movement behavior in the current time period, and determine the current state as a non-movement state; The state determination result is then output as Sta, indicating whether the current state is in motion.
7. The method for real-time emotion monitoring based on a smart bracelet according to claim 6, characterized in that: S3 includes S31; S31. Read the motion determination result Sta to determine whether the user is in motion during the current time period; When the motion determination result Sta indicates that the user is in motion, the emotion analysis process is prevented from being triggered, and an emotion triggering result Trg indicating that the emotion analysis process is prohibited from being executed is generated. When the motion determination result Sta indicates that the user is not in motion, the control emotion analysis process is allowed to be triggered, and an emotion trigger result Trg indicating that the emotion analysis process is allowed to be executed is generated.
8. A method for real-time emotion monitoring based on a smart bracelet according to claim 7, characterized in that: S4 includes S41; S41. Read the emotion triggering result Trg to determine whether to allow the emotion analysis process to be executed; When the emotion trigger result Trg indicates that the emotion analysis process is allowed, the physiological perception data collection process is initiated, and physiological perception data reflecting changes in the user's physiological state are collected through the physiological sensors built into the smart bracelet. During the collection of the physiological sensory data, the corresponding collection time information is recorded for each physiological sensory data, and the physiological sensory data is organized according to a preset time window. After completing the collection and time processing of the physiological perception data, the physiological perception data is collected in chronological order to generate the emotion analysis dataset Emo for emotion analysis processing. When the emotion trigger result Trg indicates that the emotion analysis process is not allowed, the process of collecting physiological perception data is stopped or skipped, and the emotion analysis dataset Emo is not generated.
9. A method for real-time emotion monitoring based on a smart bracelet according to claim 8, characterized in that: S5 includes S51; S51. Based on the emotion analysis dataset Emo, read the physiological perception data collected during the time period during which the emotion analysis process is allowed to be executed; Emotion feature extraction processing is performed on the physiological perception data in the emotion analysis dataset Emo to obtain emotion-related features that reflect changes in the user's physiological state; Based on the aforementioned emotion-related features, the user's current emotional state is determined to obtain a determination result representing the user's current emotional state. The determination result is then output as an emotion monitoring result Res, representing the user's current emotional state.
10. A real-time emotion monitoring system based on a smart bracelet, applied to the real-time emotion monitoring method based on a smart bracelet as described in any one of claims 1 to 9, characterized in that: It includes a behavior perception module, an activity status determination module, an emotion determination trigger module, an analysis process execution module, and an emotion processing module; The behavior perception module collects behavior perception data generated by the user while wearing the smart bracelet, and generates a set of behavior state features (Mov) representing the user's physical activity state based on the behavior perception data. The activity state determination module determines the user's current physical activity state based on the behavior state feature set Mov, and generates a motion determination result Sta indicating whether the user is in motion. The emotion determination triggering module executes emotion determination triggering control based on the motion determination result Sta, and generates an emotion triggering result Trg indicating whether the emotion analysis process is allowed. When the emotion triggering result Trg indicates that the emotion analysis process is allowed, the analysis process execution module collects physiological perception data and generates an emotion analysis dataset Emo for emotion analysis. The emotion processing module performs emotion analysis processing based on the emotion analysis dataset Emo, and outputs the emotion monitoring result Res, which represents the user's current emotional state.
