Multi-device concurrent emotion screening intelligent chip and terminal

Abstract

The invention discloses a multi-device concurrent emotion screening intelligent chip and terminal, and relates to the technical field of emotion screening, the multi-device concurrent emotion screening intelligent chip comprises a physical sign parameter acquisition module used for acquiring multi-source physical sign parameters; the data validity analysis module is used for performing consistency analysis on the multi-source physical sign parameters to obtain data validity parameters; the validity threshold acquisition module is used for acquiring a validity threshold; the acquisition frequency correction module is used for obtaining a corrected acquisition frequency to perform data acquisition to obtain an updated physical sign parameter and perform emotion recognition when the data validity parameter is smaller than a validity threshold value; and the acquisition frequency optimization module is used for performing emotion recognition on the multi-source physical sign parameters when the data validity parameters are greater than or equal to the validity threshold value, optimizing the acquisition frequency, acquiring the efficiency acquisition frequency and acquiring the multi-source physical sign parameters. The technical problem that an emotion screening data acquisition strategy in the prior art cannot adapt to an actual scene is solved.

Description

Technical Field

[0001] This invention relates to the field of emotion screening technology, specifically to a smart chip and terminal for multi-device concurrent emotion screening. Background Technology

[0002] Existing emotion screening technologies typically rely on a single type of physiological sensor or a device with a fixed sampling frequency to acquire user vital sign data. These methods often employ a static acquisition strategy, meaning that all users are sampled at a preset, uniform frequency, ignoring the dynamic changes in the quality of vital sign signals among different users and in different scenarios. This makes it difficult to achieve an adaptive balance between ensuring data validity and improving acquisition efficiency. Summary of the Invention

[0003] This application provides a multi-device concurrent emotion screening smart chip and terminal to address the technical problem in the prior art where fixed emotion screening data acquisition strategies lead to an inability to dynamically adapt data quality and acquisition efficiency.

[0004] In view of the above problems, this application provides a smart chip and terminal for multi-device concurrent emotion screening.

[0005] In a first aspect, this application provides a multi-device concurrent emotion screening smart chip, the smart chip comprising:

[0006] The vital signs parameter acquisition module is used to acquire multi-source vital signs parameters based on multi-source emotion perception devices;

[0007] The data validity analysis module is used to perform consistency analysis on the multi-source vital signs parameters and obtain data validity parameters;

[0008] The validity threshold acquisition module is used to acquire the validity threshold based on the multi-source vital sign parameters of users in the same scenario in the current scenario;

[0009] The data acquisition frequency correction module is used to correct the data acquisition frequency based on the data validity parameter when the data validity parameter is less than the validity threshold, obtain the corrected acquisition frequency for data acquisition, obtain updated vital sign parameters, and perform emotion recognition.

[0010] The acquisition frequency optimization module is used to perform emotion recognition on the multi-source vital signs parameters when the data validity parameter is greater than or equal to the validity threshold, and optimize the acquisition frequency based on the current user's emotion recognition result and the emotion recognition result set of multiple users in the current scene to obtain an efficient acquisition frequency for acquiring multi-source vital signs parameters.

[0011] Secondly, this application provides a multi-device concurrent emotion screening terminal, including:

[0012] Such as the first aspect: a smart chip for multi-device concurrent emotion screening.

[0013] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0014] This application proposes a multi-device concurrent emotion screening intelligent chip and terminal. It obtains data validity parameters by performing real-time consistency analysis on multi-source vital sign parameters and dynamically sets validity thresholds based on the vital sign data of multiple users in the current scenario. Then, based on the comparison results of the data validity parameters and the thresholds, it adaptively adjusts the acquisition frequency: frequency correction is performed to improve acquisition quality when data validity is insufficient; when data validity meets the standard, the frequency is optimized by combining individual and group emotion recognition results to reduce acquisition overhead, thus achieving a dynamic balance between data quality assurance and acquisition efficiency. Compared with traditional methods, the technical solution provided in this application significantly improves the adaptive capability of vital sign data acquisition in multi-device concurrent scenarios. It can flexibly switch between correction and optimization modes according to the signal quality fluctuations of different users and the distribution characteristics of group data within the scenario, effectively avoiding data validity fluctuations and resource waste caused by fixed acquisition strategies. This application achieves the technical effect of balancing data acquisition reliability, emotion recognition accuracy, and operational efficiency in multi-source emotion perception. Attached Figure Description

[0015] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 This is a schematic diagram of the structure of a multi-device concurrent emotion screening smart chip provided in an embodiment of this application.

[0017] Figure 2 This is a schematic diagram of the structure of a multi-device concurrent emotion screening terminal provided in an embodiment of this application.

[0018] The components represented by each number in the attached diagram are explained below:

[0019] Vital signs parameter acquisition module 100, data validity analysis module 200, validity threshold acquisition module 300, acquisition frequency correction module 400, acquisition frequency optimization module 500, multi-device concurrent emotion screening terminal 600, and multi-device concurrent emotion screening smart chip 611. Detailed Implementation

[0020] This application provides a multi-device concurrent emotion screening smart chip and terminal to address the technical problem in the prior art where fixed emotion screening data collection strategies lead to an inability to dynamically adapt data quality and collection efficiency.

[0021] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0022] It should be noted that the terms "comprising" and "having" are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or server that includes a series of steps or units is not necessarily limited to those steps or units that are explicitly listed, but may include other steps or modules that are not explicitly listed or that are inherent to these processes, methods, products, or devices.

[0023] Example 1, as Figure 1 As shown, this application provides a multi-device concurrent emotion screening smart chip, which includes:

[0024] The vital signs parameter acquisition module 100 is used to acquire multi-source vital signs parameters based on a multi-source emotion perception device.

[0025] In the process of emotion screening, it is usually necessary to rely on a single type of physiological sensor to collect user vital sign data. The parameters obtained are limited in dimension and cannot fully reflect the changes in an individual's emotional state. At the same time, when multiple emotion sensing devices collect data concurrently, there is a lack of a unified time reference between different devices, which leads to the collected parameters being out of sync in time and cannot form an effective basis for correlation analysis.

[0026] In the smart chip provided in this application embodiment, the execution steps of the vital sign parameter acquisition module 100 include:

[0027] Heart rate parameters and RRI parameters are collected by a multi-source emotion sensing device. The heart rate parameter is the number of heartbeats per unit time, and the RRI parameter is the time interval sequence between two consecutive heartbeats.

[0028] The motion sensor of the multi-source emotion sensing device collects activity parameters, which are the average acceleration of human limb movement per unit time.

[0029] The collected heart rate parameters, RRI parameters, and activity level parameters are time-stamped and linked together to form the multi-source vital signs parameters.

[0030] In this embodiment of the application, the vital signs parameter acquisition module 100 is used to acquire multi-source vital signs parameters based on a multi-source emotion perception device.

[0031] Specifically, firstly, heart rate parameters and RRI parameters are collected using a multi-source emotion sensing device. The heart rate parameter is the number of heartbeats per unit time, and the RRI parameter is the time interval sequence between two consecutive heartbeats. For example, the heart rate sensor of the multi-source emotion sensing device collects the heart rate, which is the number of heartbeats per unit time; for instance, the heart rate parameter collected at a certain moment is 75 beats per minute. The RRI parameter is collected using the electrocardiogram (ECG) sensor of the multi-source emotion sensing device. The RRI parameter is the time interval sequence between two consecutive heartbeats. The device detects the peak value of the R wave in the ECG signal and records the time interval between adjacent R waves; for example, the continuously recorded time interval sequence is 800 milliseconds, 820 milliseconds, 790 milliseconds, etc.

[0032] Furthermore, the motion sensor of the multi-source emotion sensing device collects activity parameters, which are the average acceleration of human limb movements per unit time. For example, the motion sensor of the multi-source emotion sensing device continuously collects triaxial acceleration data and calculates the average value of the composite acceleration vector per unit time; for example, the average acceleration is calculated to be 0.05g within 1 second.

[0033] Furthermore, the collected heart rate parameters, RRI parameters, and activity level parameters are time-stamped and synchronized to form the multi-source vital signs parameters. Specifically, within the same time window, the data collected by the heart rate sensor, ECG sensor, and motion sensor are marked with the same timestamp and integrated to form a structured multi-source vital signs parameter record, which includes timestamps, heart rate values, RRI sequences, and average activity levels.

[0034] By collecting heart rate parameters, RRI parameters, and activity level parameters through multi-source emotion sensing devices, and by synchronizing and binding the collected multi-source parameters with timestamps, a structured multi-source vital sign parameter was formed, realizing the alignment and fusion of physiological signals of different dimensions in the time dimension.

[0035] The data validity analysis module 200 is used to perform consistency analysis on the multi-source vital signs parameters and obtain data validity parameters.

[0036] During the acquisition of multi-source vital signs parameters, due to factors such as loose equipment, motion interference, and poor signal coupling, the correlation between data acquired from different channels often decreases. For example, the physiological correlation between heart rate parameters and ECG interval sequences weakens, or there is a mismatch between heart rate characteristics and activity level characteristics.

[0037] In the smart chip provided in this application embodiment, the execution steps of the data validity analysis module 200 include:

[0038] The correlation between heart rate and RRI parameters in multi-source vital signs parameters is calculated to obtain the correlation degree of vital signs;

[0039] Based on the heart rate parameters, an analysis and calculation are performed to obtain the heart rate stability coefficient;

[0040] Calculate the deviation of the heart rate stability coefficient and the activity level parameter to obtain the behavioral consistency parameter;

[0041] Calculate 1 and subtract the behavior consistency parameter to obtain the behavior validity parameter;

[0042] The correlation between the vital signs and the validity parameters of the behavior are weighted and calculated to obtain the data validity parameters.

[0043] In this embodiment of the application, the data validity analysis module 200 is used to perform consistency analysis on the multi-source vital signs parameters to obtain data validity parameters.

[0044] Specifically, firstly, the correlation between heart rate and RRI parameters in the multi-source vital sign parameters is calculated to obtain the vital sign correlation degree. For example, a series of continuously collected heart rate parameters and corresponding RRI parameter sequences over a period of time are extracted from the multi-source vital sign parameters. For instance, the heart rate value sequence collected within 30 seconds is 75, 76, and 74 beats per minute. Simultaneously, the RRI time interval sequence within this period is converted into a series of average RRI values ​​per minute, resulting in corresponding average RRI sequences of 800 milliseconds, 795 milliseconds, and 805 milliseconds. The Pearson correlation coefficient between the two sets of sequences is calculated for correlation analysis, yielding a vital sign correlation degree of 0.92.

[0045] Furthermore, based on the heart rate parameters, analysis and calculation are performed to obtain the heart rate stability coefficient. For example, heart rate values ​​at multiple consecutive time points are extracted from multi-source vital signs parameters. For instance, heart rate is collected once per second for 10 consecutive seconds, resulting in a heart rate sequence of 75, 76, 75, 77, and 75 beats per minute. The mean of this sequence is calculated to be approximately 75.6, and the standard deviation is calculated to be 0.8. Dividing the standard deviation by the mean yields a coefficient of variation of approximately 0.01. Subtracting this coefficient of variation from 1 gives a heart rate stability coefficient of approximately 0.99.

[0046] Further, the deviation of the heart rate stability coefficient and the activity level parameter is calculated to obtain the behavioral consistency parameter. For example, the mean activity acceleration per second over a continuous 10-second period is 0.05g, 0.06g, 0.05g, 0.07g, and 0.05g. The mean of this sequence is calculated to be 0.056, and the standard deviation is calculated to be 0.008. Dividing the standard deviation by the mean yields an activity level fluctuation coefficient of approximately 0.14. Subtracting this fluctuation coefficient from 1 yields an activity level stability coefficient of approximately 0.86. The absolute value of the difference between the heart rate stability coefficient and the activity level stability coefficient is calculated to obtain the behavioral consistency parameter: |0.86 - 0.99| = 0.13.

[0047] Further, subtract the behavioral consistency parameter from 1 to obtain the behavioral effectiveness parameter. For example, if the behavioral consistency parameter is 0.13, then the behavioral effectiveness parameter = 1 - 0.13 = 0.87. A larger value indicates a higher degree of effective synchronization between heart rate and activity level.

[0048] Furthermore, the correlation between the vital signs and the behavioral validity parameters are weighted and calculated to obtain the data validity parameter. For example, if behavioral validity is more important in the current analysis, the weight of the behavioral validity parameter can be set to 0.6, and the weight of the vital sign correlation parameter can be set to 0.4. The calculated data validity parameter is: 0.4 × vital sign correlation parameter + 0.6 × behavioral validity parameter = 0.4 × 0.92 + 0.6 × 0.87 = 0.89.

[0049] By calculating the correlation between heart rate and RRI parameters among multi-source vital signs, the correlation degree of vital signs is obtained. Then, the behavioral consistency parameter is calculated by combining the heart rate stability coefficient and activity level parameter. Finally, the data validity parameter is obtained by weighted calculation, thus realizing the quantitative assessment of the internal consistency of multi-source vital signs parameters.

[0050] The validity threshold acquisition module 300 is used to acquire the validity threshold based on the multi-source vital sign parameters of users in the same scenario in the current scenario.

[0051] In multi-device concurrent emotion screening scenarios, there are significant differences in the quality of vital signs signals among different users and even among the same user in different scenarios. Existing technologies typically use fixed preset thresholds to determine data validity, which cannot adapt to changes in scenarios and group differences. For example, in crowded scenarios, the distribution of signal quality among individuals may exhibit large dispersion, while in quiet scenarios it is relatively concentrated. Fixed thresholds often fail to meet the actual needs of different scenarios.

[0052] In the smart chip provided in this application embodiment, the execution steps of the validity threshold acquisition module 300 include:

[0053] Obtain multi-source vital signs parameters of all users in the same scene within the current scene to form a set of scene data validity parameters;

[0054] Calculate the arithmetic mean of the set of scene data validity parameters to obtain the scene validity mean;

[0055] Calculate the coefficient of variation of the set of scene data validity parameters to obtain the scene validity dispersion;

[0056] The validity threshold is obtained by weighting the mean of the scene validity and the dispersion of the scene validity.

[0057] In this embodiment of the application, the validity threshold acquisition module 300 is used to acquire the validity threshold based on the multi-source vital sign parameters of users in the same scenario in the current scenario.

[0058] Specifically, firstly, multi-source vital sign parameters of all users in the same scenario are acquired to form a set of scenario data validity parameters. For example, firstly, multi-source vital sign parameters of all users in the same scenario are acquired, and then consistency analysis is performed on the multi-source vital sign parameters of each user to form a set of scenario data validity parameters. Specifically, in a scenario where multiple people are simultaneously undergoing emotion screening, such as 10 users in the same meeting room, each user wears a multi-source emotion sensing device that collects their own multi-source vital sign parameters. The data validity analysis module 200 calculates the data for each user sequentially to obtain the corresponding data validity parameters for each user. These parameters are then collected together to form a set of scenario data validity parameters. For example, the values ​​in the set of scenario data validity parameters might be 0.85, 0.78, 0.92, 0.81, 0.88, 0.79, 0.90, 0.83, 0.86, and 0.84.

[0059] Further, the arithmetic mean of the set of scene data validity parameters is calculated to obtain the scene validity mean. For example, the scene validity mean = (0.85+0.78+0.92+0.81+0.88+0.79+0.90+0.83+0.86+0.84) / 10 = 0.846.

[0060] Further, the coefficient of variation of the set of scene data validity parameters is calculated to obtain the scene validity dispersion. Specifically, the standard deviation of the set of scene data validity parameters is first calculated, for example, 0.044. Then, this standard deviation is divided by the scene validity mean of 0.846, resulting in a coefficient of variation of approximately 0.052, which is the scene validity dispersion. The larger the scene validity dispersion value, the greater the relative difference between the various user data validity parameters within the scene.

[0061] Furthermore, a validity threshold is obtained by performing a weighted calculation based on the mean and dispersion of scene validity. For example, the mean of scene validity reflects the overall situation within the scene and is given a relatively large weight, such as 0.7 for the mean and 1-0.7=0.3 for the dispersion. The validity threshold is then calculated as 0.846×0.7+0.052×0.3=0.61. This validity threshold is used for subsequent comparison with the data validity parameters of an individual user to determine whether the quality of the data collected by that user meets the validity standard within the scene.

[0062] By acquiring multi-source vital signs parameters of all users in the same scenario and performing consistency analysis, a set of scenario data validity parameters is formed. Then, the mean and standard deviation of scenario validity are calculated, and a validity threshold is obtained by weighted calculation, thus realizing the scenario adaptive setting of the judgment benchmark.

[0063] The acquisition frequency correction module 400 is used to correct the acquisition frequency based on the data validity parameter when the data validity parameter is less than the validity threshold, acquire the corrected acquisition frequency for data acquisition, acquire updated vital sign parameters, and perform emotion recognition.

[0064] When the data validity parameter is lower than the judgment standard, it indicates that the internal consistency of the currently collected multi-source vital signs parameters is poor. If data collection continues at the original collection frequency, low-quality data will continue to be acquired, which will not effectively improve the signal quality. As a result, the emotion recognition process will always be based on unreliable data.

[0065] In the smart chip provided in this application embodiment, the execution steps of the acquisition frequency correction module 400 include:

[0066] When the data validity parameter is less than the validity threshold, the invalidity reason is determined and the invalidity reason is obtained;

[0067] When the data validity parameter is less than the validity threshold, an invalidity reason determination is performed to obtain the invalidity reason, including:

[0068] When the data validity parameter is less than the validity threshold, an invalidity reason determination is performed;

[0069] If the correlation between vital signs and physical signs is lower than the correlation threshold and the validity parameter of behavior is less than the validity threshold of behavior, the reason for invalidity is that the correlation between heart rate and RRI parameter acquisition is abnormal. The waveform peak deviation rate of heart rate parameter and the peak missing rate of RRI are calculated by weighting, and the waveform abnormality correction coefficient is obtained.

[0070] If the behavior validity parameter is greater than or equal to the behavior validity threshold and the vital sign correlation degree is greater than or equal to the vital sign correlation threshold, the invalidity is determined to be due to abnormal matching between behavior and heart rate characteristics. The ratio of the signal fluctuation coefficient of the heart rate parameter to the signal fluctuation coefficient of the activity level is calculated to obtain the matching abnormality correction coefficient.

[0071] If the correlation degree of vital signs is lower than the correlation threshold of vital signs and the validity parameter of behavior is greater than or equal to the validity threshold of behavior, the invalidity is determined to be due to double anomaly in correlation matching. The waveform anomaly correction coefficient and the matching anomaly correction coefficient are calculated by weighting to obtain the double anomaly correction coefficient.

[0072] The anomaly correction coefficient corresponding to the invalidation cause is used as the frequency correction coefficient;

[0073] Based on the aforementioned reasons for invalidity, obtain the frequency correction coefficient;

[0074] The corrected sampling frequency is obtained by adding 1 to the frequency correction factor and multiplying it by the current sampling frequency;

[0075] Data is collected using the corrected acquisition frequency to obtain updated vital sign parameters, and consistency analysis and iterative updates are continued until the data validity parameters are greater than or equal to the validity threshold.

[0076] The updated vital signs parameters are input into the emotion recognition analyzer to obtain the emotion recognition results;

[0077] The construction of the emotion recognition analyzer includes:

[0078] Obtain multi-source vital signs parameters and corresponding emotion labels of the samples as the sample training set;

[0079] An emotion recognition analyzer was built based on neural networks.

[0080] The emotion recognition analyzer is trained using the sample training set until convergence, thus obtaining the trained emotion recognition analyzer.

[0081] In this embodiment of the application, the acquisition frequency correction module 400 is used to correct the acquisition frequency based on the data validity parameter when the data validity parameter is less than the validity threshold, acquire the corrected acquisition frequency for data acquisition, acquire updated vital sign parameters, and perform emotion recognition.

[0082] Specifically, firstly, when the data validity parameter is less than the validity threshold, an invalidity reason determination is performed to obtain the invalidity reason. For example, a user's data validity parameter is 0.58, while the validity threshold is 0.61. Since the data validity parameter is less than the validity threshold, an invalidity reason determination is required. The user's vital sign correlation score is obtained as 0.65, and the behavioral consistency parameter as 0.88. For instance, based on the average vital sign correlation score and the average behavioral validity parameter of users in the same scenario, the vital sign correlation threshold is obtained as 0.70, and the behavioral validity threshold as 0.80.

[0083] Furthermore, if the correlation between vital signs and the behavior validity parameter is lower than the threshold, the invalidity is determined to be due to abnormal correlation between heart rate and RRI parameter acquisition. The waveform peak deviation rate and RRI peak missing rate of the heart rate parameter are calculated using a weighted average, and a waveform abnormality correction coefficient is obtained. The waveform peak deviation rate of the heart rate parameter refers to the difference between the actual detected peak amplitude and the theoretically expected peak amplitude in continuously acquired heart rate waveforms, where the waveform peak point for each heartbeat cycle is detected. For example, this includes the proportion of actual peaks deviating from the normal shape due to motion interference or poor equipment coupling causing waveform distortion. This deviation rate is between 0 and 1; a larger value indicates more severe waveform distortion. The RRI peak missing rate refers to the proportion of missing R waves in the total number of R waves that should be detected, where some R wave peaks cannot be correctly identified and extracted due to signal quality degradation during ECG R wave detection. This missing rate is between 0 and 1; a larger value indicates more severe R wave missed detection. For example, in a 10-second ECG signal, theoretically there should be 10 R wave peaks, but only 8 are actually detected, resulting in a missing rate of 0.20. The waveform peak deviation rate and the RRI peak missing rate are weighted and calculated. For example, if the deviation rate weight is set to 0.5 and the missing rate weight is 0.5, and the waveform peak deviation rate is 0.15 and the RRI peak missing rate is 0.20, the weighted calculation yields a waveform anomaly correction coefficient of 0.18.

[0084] Furthermore, if the behavioral validity parameter is greater than or equal to the behavioral validity threshold and the vital sign correlation is greater than or equal to the vital sign correlation threshold, the invalidity is determined to be due to an abnormal match between the behavior and heart rate characteristics. The ratio of the signal fluctuation coefficient of the heart rate parameter to the signal fluctuation coefficient of the activity level is calculated to obtain a matching anomaly correction coefficient. Specifically, the signal fluctuation coefficient of the heart rate parameter is calculated by dividing the average peak amplitude of the heart rate waveform by the overall average amplitude of the waveform within a unit of time. This ratio reflects the prominence of the heart rate signal peak relative to the overall amplitude; a larger value indicates a more pronounced peak feature in the heart rate signal. The signal fluctuation coefficient of the activity level parameter is calculated by dividing the average peak amplitude of the activity level waveform by the overall average amplitude of the waveform within the same unit of time. This ratio reflects the prominence of the activity level signal peak relative to the overall amplitude. For example, if the average peak amplitude of the heart rate waveform is 0.8 mV and the overall average amplitude is 0.4 mV within 1 second, the heart rate signal fluctuation coefficient is 2.0; if the average peak amplitude of the activity level waveform is 0.3g and the overall average amplitude is 0.2g, the activity level signal fluctuation coefficient is 1.5. The ratio of the heart rate signal fluctuation coefficient to the activity level signal fluctuation coefficient was calculated, and the matching anomaly correction coefficient was obtained as 2 / 1.5 = 1.33. This ratio is greater than 1, indicating that the peak characteristics of the heart rate signal are more prominent than those of the activity level.

[0085] Furthermore, if the correlation degree of the vital signs is lower than the correlation threshold and the behavior validity parameter is greater than or equal to the behavior validity threshold, the invalidity is determined to be due to double anomaly in the correlation matching. The waveform anomaly correction coefficient and the matching anomaly correction coefficient are then calculated using weighted averages to obtain the double anomaly correction coefficient. For example, the weights of both the waveform anomaly correction coefficient and the matching anomaly correction coefficient are set to 0.5 to comprehensively consider the impact of correlation matching. The calculated double anomaly correction coefficient is: 0.5 × waveform anomaly correction coefficient + 0.5 × matching anomaly correction coefficient = 0.5 × 0.18 + 0.5 × 1.33 = 0.74.

[0086] Furthermore, the anomaly correction coefficient corresponding to the invalidation cause is used as the frequency correction coefficient. If the invalidation cause is a double anomaly in the correlation matching, then the frequency correction coefficient = double anomaly correction parameter = 0.74.

[0087] Furthermore, the corrected sampling frequency is obtained by adding 1 to the frequency correction factor and multiplying by the current sampling frequency. Corrected sampling frequency = (1 + frequency correction factor) × current sampling frequency. For example, if the current sampling frequency is 5 times per second, then the corrected sampling frequency = (1 + 0.74) × 5 = 8.7 times per second. If the result is not an integer, it is rounded up to obtain a sampling frequency of 9 times per second.

[0088] Furthermore, data is collected using the corrected acquisition frequency to obtain updated vital sign parameters, and consistency analysis and iterative updates continue until the data validity parameter is greater than or equal to the validity threshold. Specifically, the acquisition frequency is adjusted to 9 times per second, and multi-source vital sign parameters are re-acquired. The data validity analysis module 200 calculates new data validity parameters. If the new data validity parameter is still less than the validity threshold, the above invalidity cause determination and frequency correction process is repeated until the data validity parameter is greater than or equal to the validity threshold.

[0089] Furthermore, the updated vital signs parameters are input into the emotion recognition analyzer to obtain the emotion recognition results.

[0090] The construction of the emotion recognition analyzer includes:

[0091] First, obtain the multi-source vital signs parameters and corresponding emotion labels of the samples as the sample training set. For example, collect 10,000 historically collected multi-source vital signs parameter records, each containing heart rate parameters, RRI parameters, and activity level parameters, and have professionals label them with corresponding emotion recognition results, such as calm, mild pleasure, moderate pleasure, mild tension, moderate tension, mild fatigue, moderate fatigue, etc., and integrate them to obtain the sample training set.

[0092] Furthermore, an emotion recognition analyzer is constructed based on a neural network. For example, a three-layer fully connected neural network structure is adopted. The input layer has 128 neurons, corresponding to the multi-source vital sign parameter feature dimensions of the input. The first hidden layer has 64 neurons, using a linear rectified function as the activation function. The second hidden layer has 32 neurons, also using a linear rectified function as the activation function. The output layer has 1 neuron, using a normalized exponential function as the activation function for multi-class classification output.

[0093] Further, the emotion recognition analyzer is trained using the sample training set until convergence, resulting in a trained emotion recognition analyzer. For example, the emotion recognition analyzer is trained using the sample training set, the cross-entropy loss function is used to calculate the loss between the predicted result and the true label, an adaptive moment estimation optimizer is used to update the parameters, the learning rate is set to 0.001, the batch size is 32, and the training is iterated for 200 rounds until the loss function converges, resulting in a trained emotion recognition analyzer.

[0094] When the data validity parameter is less than the validity threshold, the acquisition frequency is corrected based on the data validity parameter. Data is then acquired at the corrected acquisition frequency to obtain updated vital sign parameters until the data validity parameter meets the requirements, at which point emotion recognition is performed. This application can proactively adjust the acquisition strategy when data consistency is insufficient, improve signal acquisition quality through frequency correction, and ensure that subsequently acquired vital sign parameters have higher validity, thereby providing reliable data input for emotion recognition.

[0095] The acquisition frequency optimization module 500 is used to perform emotion recognition on the multi-source vital signs parameters when the data validity parameter is greater than or equal to the validity threshold, and optimize the acquisition frequency based on the current user's emotion recognition result and the emotion recognition result set of multiple users in the current scene to obtain an efficient acquisition frequency for acquiring multi-source vital signs parameters.

[0096] When the data validity parameters meet the judgment criteria, it indicates that the currently collected multi-source vital signs parameters have good consistency and can be used for emotion recognition. However, while ensuring data quality, existing technologies usually maintain a fixed high-frequency collection pattern, which may lead to wasted resources.

[0097] In the smart chip provided in this application embodiment, the execution steps of the acquisition frequency optimization module 500 include:

[0098] When the data validity parameter is greater than or equal to the validity threshold, the multi-source vital sign parameters are input into the emotion recognition analyzer to obtain the emotion recognition result;

[0099] Based on the emotion recognition results, efficiency optimization parameters are obtained;

[0100] Among them, efficiency optimization parameters are obtained based on the emotion recognition results, including:

[0101] Obtain the emotion category corresponding to the current user's emotion recognition result;

[0102] Perform emotion recognition on multiple users in the same scene, obtain an emotion recognition result set, classify the emotions, and obtain a group emotion profile;

[0103] Based on the aforementioned group emotional profile, obtain the group efficiency optimization coefficient;

[0104] Based on the current user's emotion classification, obtain the user efficiency optimization coefficient;

[0105] The efficiency optimization coefficient is obtained by weighting the group efficiency optimization coefficient and the user efficiency optimization coefficient.

[0106] The efficiency acquisition frequency is obtained by multiplying the efficiency optimization coefficient by the acquisition frequency, and multi-source vital signs parameters are acquired.

[0107] In this embodiment of the application, the acquisition frequency optimization module 500 is used to perform emotion recognition on the multi-source vital signs parameters when the data validity parameter is greater than or equal to the validity threshold, and optimize the acquisition frequency based on the current user's emotion recognition result and the emotion recognition result set of multiple users in the current scene to obtain an efficient acquisition frequency for acquiring multi-source vital signs parameters.

[0108] Specifically, firstly, when the data validity parameter is greater than or equal to the validity threshold, the multi-source vital signs parameters are input into the emotion recognition analyzer to obtain the emotion recognition result. For example, if a user's multi-source vital signs parameters are calculated by the data validity analysis module 200 to have a data validity parameter of 0.86, which is higher than the validity threshold of 0.61, indicating that the quality of the currently collected data is good, this set of parameters is input into the pre-trained emotion recognition analyzer, which outputs that the user's emotion is classified as calm.

[0109] Furthermore, emotion recognition is performed on multiple users in the same scene to obtain an emotion recognition result set, and then the emotions are categorized to obtain a group emotion profile. For example, the emotion recognition result for the current user corresponds to the emotion category of "calm." Simultaneously, emotion recognition is performed on multiple users in the same scene separately to obtain an emotion recognition result set, and these results are statistically categorized to obtain a group emotion profile. For instance, if 10 users in the same meeting room are identified by the emotion recognition analyzer, their emotion categories are: 5 calm, 2 slightly happy, 2 slightly tense, and 1 slightly tired. This group emotion profile indicates that most users in this scene are in a calm state. Based on the group emotion profile, a group efficiency optimization coefficient is obtained. When the proportion of calm states in the group exceeds 50%, the group efficiency optimization coefficient is set to 0.7, indicating that the overall emotion in the scene is relatively stable, and the collection frequency can be appropriately reduced. When other emotion categories account for more than 50%, the group efficiency optimization coefficient can be obtained based on the proportion of data for that emotion category in the sample training set. For example, if the proportion of data for the fatigue category in the sample training set is 0.1, then the group efficiency optimization coefficient can be set to 1-0.1=0.9.

[0110] Furthermore, based on the current user's emotion classification, a user efficiency optimization coefficient is obtained. For example, a pre-established correspondence between emotion classification and user efficiency optimization coefficient is established: the user efficiency optimization coefficient for a calm state is 0.7, for a pleasant emotion classification it is 0.8, for a tense emotion classification it is 1.0, and for a fatigued emotion classification it is 0.9. Based on the current user's emotion classification, the user efficiency optimization coefficient is obtained. Since the current user is in a calm state, the corresponding user efficiency optimization coefficient is 0.7, indicating that the user's emotions are stable and their vital signs change slowly, allowing for a more appropriate slowdown in the data collection frequency.

[0111] Further, the group efficiency optimization coefficient and the user efficiency optimization coefficient are weighted and calculated to obtain the overall efficiency optimization coefficient. For example, if the current group has a small number of members, the weight of the group efficiency optimization coefficient is set to 0.4, then the weight of the user efficiency optimization coefficient is 1 - 0.4 = 0.6. The calculated efficiency optimization coefficient is: 0.4 × group efficiency optimization coefficient + 0.6 × user efficiency optimization coefficient = 0.4 × 0.9 + 0.6 × 0.7 = 0.78.

[0112] Furthermore, the efficiency optimization coefficient is multiplied by the acquisition frequency to obtain the efficient acquisition frequency for acquiring multi-source vital signs parameters. For example, if the current acquisition frequency is 10 acquisitions per second, multiplying by the efficiency optimization coefficient 0.78 and rounding up if the result is not an integer, the efficient acquisition frequency is obtained as 8 acquisitions per second. Since the current data quality meets the requirements and the emotional states of individuals and groups are relatively stable, reducing the acquisition frequency from 10 acquisitions per second to 8 acquisitions per second can reduce the system's energy consumption and data processing burden while maintaining the accuracy of emotion recognition. Subsequent acquisitions and verification of multi-source vital signs parameters will continue according to this efficient acquisition frequency.

[0113] When the data validity parameter is greater than or equal to the validity threshold, emotion recognition is performed on multi-source vital signs parameters. Based on the current user's emotion recognition result and the emotion recognition result set of multiple users in the current scene, the collection frequency is optimized. The efficient collection frequency is used for subsequent data collection, which improves the collection efficiency under high data quality. Under the premise of ensuring data validity, the collection frequency can be dynamically adjusted according to individual emotional characteristics and group emotional profiles. While meeting the recognition requirements, unnecessary collection overhead is reduced, achieving a coordinated unity between data quality assurance and system energy efficiency optimization.

[0114] Example 2, as Figure 2 As shown, this embodiment of the invention also provides a multi-device concurrent emotion screening terminal 600, comprising:

[0115] As exemplified in Example 1, a multi-device concurrent emotion screening smart chip.

[0116] In summary, the embodiments of this application have at least the following technical effects:

[0117] This application proposes a multi-device concurrent emotion screening intelligent chip and terminal. It obtains data validity parameters by performing real-time consistency analysis on multi-source vital sign parameters and dynamically sets validity thresholds based on the vital sign data of multiple users in the current scenario. Then, based on the comparison results of the data validity parameters and the thresholds, it adaptively adjusts the acquisition frequency: frequency correction is performed to improve acquisition quality when data validity is insufficient; when data validity meets the standard, the frequency is optimized by combining individual and group emotion recognition results to reduce acquisition overhead, thus achieving a dynamic balance between data quality assurance and acquisition efficiency. Compared with traditional methods, the technical solution provided in this application significantly improves the adaptive capability of vital sign data acquisition in multi-device concurrent scenarios. It can flexibly switch between correction and optimization modes according to the signal quality fluctuations of different users and the distribution characteristics of group data within the scenario, effectively avoiding data validity fluctuations and resource waste caused by fixed acquisition strategies. This application achieves the technical effect of balancing data acquisition reliability, emotion recognition accuracy, and operational efficiency in multi-source emotion perception.

[0118] It should be noted that the order of the embodiments described above is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. Furthermore, the above description focuses on specific embodiments of this specification. Additionally, the processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some implementations, multitasking and parallel processing are possible or may be advantageous.

[0119] The above description is only a preferred embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

[0120] This specification and accompanying drawings are merely illustrative examples of this application and are intended to cover any and all modifications, variations, combinations, or equivalents within the scope of this application. Clearly, those skilled in the art can make various alterations and modifications to this application without departing from its scope. Therefore, if such modifications and modifications fall within the scope of this application and its equivalents, this application intends to include such modifications and modifications.

Claims

1. A smart chip for multi-device concurrent emotion screening, characterized in that, include: The vital signs parameter acquisition module is used to acquire multi-source vital signs parameters based on multi-source emotion perception devices; The data validity analysis module is used to perform consistency analysis on the multi-source vital signs parameters and obtain data validity parameters; The validity threshold acquisition module is used to acquire the validity threshold based on the multi-source vital sign parameters of users in the same scenario in the current scenario; The data acquisition frequency correction module is used to correct the data acquisition frequency based on the data validity parameter when the data validity parameter is less than the validity threshold, obtain the corrected acquisition frequency for data acquisition, obtain updated vital sign parameters, and perform emotion recognition. The acquisition frequency optimization module is used to perform emotion recognition on the multi-source vital signs parameters when the data validity parameter is greater than or equal to the validity threshold, and optimize the acquisition frequency based on the current user's emotion recognition result and the emotion recognition result set of multiple users in the current scene to obtain an efficient acquisition frequency for acquiring multi-source vital signs parameters.

2. The multi-device concurrent emotion screening smart chip according to claim 1, characterized in that, Based on multi-source emotion perception devices, multi-source vital sign parameters are acquired, including: Heart rate parameters and RRI parameters are collected by a multi-source emotion sensing device. The heart rate parameter is the number of heartbeats per unit time, and the RRI parameter is the time interval sequence between two consecutive heartbeats. The motion sensor of the multi-source emotion sensing device collects activity parameters, which are the average acceleration of human limb movement per unit time. The collected heart rate parameters, RRI parameters, and activity level parameters are time-stamped and linked together to form the multi-source vital signs parameters.

3. The multi-device concurrent emotion screening smart chip according to claim 1, characterized in that, Consistency analysis was performed on the multi-source vital signs parameters to obtain data validity parameters, including: The correlation between heart rate and RRI parameters in multi-source vital signs parameters is calculated to obtain the correlation degree of vital signs; Based on the heart rate parameters, an analysis and calculation are performed to obtain the heart rate stability coefficient; Calculate the deviation of the heart rate stability coefficient and the activity level parameter to obtain the behavioral consistency parameter; Calculate 1 and subtract the behavior consistency parameter to obtain the behavior validity parameter; The correlation between the vital signs and the validity parameters of the behavior are weighted and calculated to obtain the data validity parameters.

4. The multi-device concurrent emotion screening smart chip according to claim 1, characterized in that, Based on multi-source vital sign parameters of users in the same current scenario, obtain validity thresholds, including: Obtain multi-source vital signs parameters of all users in the same scene within the current scene to form a set of scene data validity parameters; Calculate the arithmetic mean of the set of scene data validity parameters to obtain the scene validity mean; Calculate the coefficient of variation of the set of scene data validity parameters to obtain the scene validity dispersion; The validity threshold is obtained by weighting the mean of the scene validity and the dispersion of the scene validity.

5. The multi-device concurrent emotion screening smart chip according to claim 1, characterized in that, When the data validity parameter is less than the validity threshold, the data collection frequency is corrected based on the data validity parameter, the corrected collection frequency is obtained for data collection, updated vital sign parameters are obtained, and emotion recognition is performed, including: When the data validity parameter is less than the validity threshold, the invalidity reason is determined and the invalidity reason is obtained; Based on the aforementioned reasons for invalidity, obtain the frequency correction coefficient; The corrected sampling frequency is obtained by adding 1 to the frequency correction factor and multiplying it by the current sampling frequency; Data is collected using the corrected acquisition frequency to obtain updated vital sign parameters, and consistency analysis and iterative updates are continued until the data validity parameters are greater than or equal to the validity threshold. The updated vital signs parameters are input into the emotion recognition analyzer to obtain the emotion recognition results.

6. The multi-device concurrent emotion screening smart chip according to claim 5, characterized in that, When the data validity parameter is less than the validity threshold, an invalidity reason determination is performed to obtain the invalidity reason, including: When the data validity parameter is less than the validity threshold, an invalidity reason determination is performed; If the correlation between vital signs and physical signs is lower than the correlation threshold and the validity parameter of behavior is less than the validity threshold of behavior, the reason for invalidity is that the correlation between heart rate and RRI parameter acquisition is abnormal. The waveform peak deviation rate of heart rate parameter and the peak missing rate of RRI are calculated by weighting, and the waveform abnormality correction coefficient is obtained. If the behavior validity parameter is greater than or equal to the behavior validity threshold and the vital sign correlation degree is greater than or equal to the vital sign correlation threshold, the invalidity is determined to be due to abnormal matching between behavior and heart rate characteristics. The ratio of the signal fluctuation coefficient of the heart rate parameter to the signal fluctuation coefficient of the activity level is calculated to obtain the matching abnormality correction coefficient. If the correlation degree of vital signs is lower than the correlation threshold of vital signs and the validity parameter of behavior is greater than or equal to the validity threshold of behavior, the invalidity is determined to be due to double anomaly in correlation matching. The waveform anomaly correction coefficient and the matching anomaly correction coefficient are calculated by weighting to obtain the double anomaly correction coefficient. The anomaly correction coefficient corresponding to the invalidation cause is used as the frequency correction coefficient.

7. The multi-device concurrent emotion screening smart chip according to claim 6, characterized in that, The construction of the emotion recognition analyzer includes: Obtain multi-source vital signs parameters and corresponding emotion labels of the samples as the sample training set; An emotion recognition analyzer was built based on neural networks. The emotion recognition analyzer is trained using the sample training set until convergence, thus obtaining the trained emotion recognition analyzer.

8. The multi-device concurrent emotion screening smart chip according to claim 1, characterized in that, When the data validity parameter is greater than or equal to the validity threshold, emotion recognition is performed on the multi-source vital signs parameters. Based on the current user's emotion recognition result and the emotion recognition result set of multiple users in the current scene, the acquisition frequency is optimized to obtain an efficient acquisition frequency for acquiring multi-source vital signs parameters, including: When the data validity parameter is greater than or equal to the validity threshold, the multi-source vital sign parameters are input into the emotion recognition analyzer to obtain the emotion recognition result; Based on the emotion recognition results, efficiency optimization parameters are obtained; The efficiency acquisition frequency is obtained by multiplying the efficiency optimization coefficient by the acquisition frequency, and multi-source vital signs parameters are acquired.

9. A multi-device concurrent emotion screening smart chip according to claim 8, characterized in that, Based on the emotion recognition results, efficiency optimization parameters are obtained, including: Obtain the emotion category corresponding to the current user's emotion recognition result; Perform emotion recognition on multiple users in the same scene, obtain an emotion recognition result set, classify the emotions, and obtain a group emotion profile; Based on the aforementioned group emotional profile, obtain the group efficiency optimization coefficient; Based on the current user's emotion classification, obtain the user efficiency optimization coefficient; The efficiency optimization coefficient is obtained by weighting the group efficiency optimization coefficient and the user efficiency optimization coefficient.

10. A multi-device concurrent emotion screening terminal, characterized in that, include: A multi-device concurrent emotion screening smart chip according to any one of claims 1-9.

Images