Distributed family health monitoring Internet of Things system based on edge computing

Abstract

The embodiment of the invention provides a distributed family health monitoring Internet of Things system based on edge computing. The method is applied to the technical field of civil blasting. The system comprises a data acquisition module used for acquiring real-time environment data, real-time behavior data and real-time physiological parameters; the behavior judgment module is used for judging the behavior type of the user according to the real-time behavior data and calculating an abnormal score; the behavior correction module is used for judging an emotional state according to the real-time physiological parameters and the real-time behavior data, correcting an abnormal score according to the emotional state, calculating environment deviation according to the real-time environment data, and compensating the abnormal score according to the environment deviation to obtain a second behavior score; and the comprehensive judgment module is used for presetting a decision level and adopting different response methods according to the second behavior score. Through cooperative work of the environment acquisition unit, the behavior acquisition unit and the wearable device, real-time acquisition and fusion of physiological, behavior and environment data are realized, and a multi-dimensional health monitoring network is constructed.

Description

Technical Field

[0001] The present disclosure relates to the field of health monitoring technology, and in particular to a distributed home health monitoring Internet of Things system based on edge computing. Background Art

[0002] The high prevalence of chronic diseases is driving the development of refined, real-time family health management. Traditional medical models, due to issues like long intervals between visits and inconvenient access to medical care, struggle to meet the needs of daily health monitoring. IoT technology, through device interconnection and data exchange, offers a new solution for family health monitoring, enabling real-time collection and analysis of physiological parameters, behavioral data, and environmental data.

[0003] IoT technology builds a home health monitoring network through wearable devices and environmental sensors. Edge computing brings data processing capabilities down to the device side, reducing data transmission latency, lowering reliance on the cloud, and improving real-time response efficiency, making it particularly suitable for resource-constrained home scenarios.

[0004] Behavioral recognition technology has been widely used in fields such as elderly care. It requires analyzing daily behavior patterns to provide early warnings of health risks. However, existing technologies are often affected by the environment, user emotions, and other factors, leading to misjudgments and false alarms. Summary of the Invention

[0005] The present disclosure provides a distributed home health monitoring Internet of Things system based on edge computing, which solves the misjudgment of slow movements caused by aging and misjudgment caused by emotions, environment, etc., and improves monitoring accuracy.

[0006] According to a first aspect of the present disclosure, a distributed home health monitoring IoT system based on edge computing is provided. The system includes: Data collection module: used to collect real-time environmental data through the environmental collection unit set in the user's home, collect the user's real-time behavior data through the behavior collection unit set by the user, and collect the user's real-time physiological parameters through the wearable device worn by the user; Behavior judgment module: used to judge the user's behavior type based on real-time behavior data and calculate anomaly scores based on the behavior type; Behavior correction module: used to determine the emotional state based on real-time physiological parameters and real-time behavioral data, correct the abnormal score based on the emotional state to obtain a first behavior score; calculate the environmental deviation based on real-time environmental data, compensate the first behavior score based on the environmental deviation, and obtain a second behavior score; Comprehensive judgment module: used to preset decision levels in the cloud and adopt different response methods according to the decision level corresponding to the user's second behavior score.

[0007] According to the above aspects and any possible implementation, there is further provided an implementation, wherein the environment acquisition unit includes a temperature and humidity sensor; The behavior collection unit includes a door magnetic sensor, a mattress pressure sensor, a millimeter wave radar and a microphone array; The wearable device is provided with a heart rate sensor and a skin conductance sensor.

[0008] According to the above aspects and any possible implementation, a further implementation is provided, wherein the method for determining the user's behavior type based on real-time physiological parameters and calculating an abnormality score based on the behavior type includes: Generate behavior templates corresponding to different behavior types through the historical behavior data collected by the behavior collection unit; the behavior templates include normal templates, risk templates and fatigue templates; Determine the corresponding real-time behavior type based on the real-time behavior data collected by the behavior collection unit, and calculate the minimum distance between the real-time behavior data and the behavior template of the corresponding real-time behavior type; Query the constructed historical transition probability matrix and calculate the reasonable value of the real-time behavior type; The anomaly score is calculated based on the behavior rationality value and the minimum distance.

[0009] According to the aspects and any possible implementations described above, an implementation is further provided, wherein the normal template, risk template, and fatigue template all include behavior duration, maximum movement speed, and joint angle variance of several types of user behaviors.

[0010] According to the above aspects and any possible implementation, an implementation is further provided, wherein the historical transition probability matrix is ​​obtained by: Calculate the conditional probability of converting from a real-time behavior type to the next behavior type, and calculate the average time required for the conversion. Determine the spatial conversion based on the data collected by the door magnetic sensor. The real-time behavior type, next behavior type, conditional probability, average time and space conversion are respectively used as columns, and various conversion situations are recorded in rows to obtain the historical transfer probability matrix.

[0011] According to the above aspects and any possible implementation, a further implementation is provided, which queries the constructed historical transition probability matrix and calculates the reasonable behavior value of the real-time behavior type, including: After identifying the user's real-time behavior type, various conversion situations corresponding to the current behavior type consistent with the user's real-time behavior type are obtained from the historical transition probability matrix, and then waiting for the next behavior type to occur; Check whether the next behavior type meets the time and space constraints, and calculate the reasonable value of the real-time behavior type.

[0012] According to the above aspects and any possible implementation, an implementation is further provided, wherein checking whether the next behavior type meets the spatiotemporal constraint conditions includes: First determine the rationality of space, then determine the rationality of time, and calculate the rationality value of behavior based on the product of the rationality of time and the rationality of space.

[0013] According to the above aspects and any possible implementation, a further implementation is provided for determining the emotional state based on real-time physiological parameters and real-time behavioral data, including: Extract features from real-time physiological parameters and real-time behavioral data, concatenate the extracted features, and output the user's emotional state after passing through a lightweight convolutional neural network; The lightweight convolutional neural network includes: a 3×1 convolutional layer, a maximum pooling layer, a 5×1 convolutional layer, a global pooling layer and a fully connected layer; The emotional states include negative emotions, positive emotions, and calm emotions.

[0014] According to the above aspects and any possible implementation, a further implementation is provided, wherein an environmental deviation is calculated based on real-time environmental data, and the first behavior score is compensated based on the environmental deviation, including: A reference temperature is calculated based on the temperature collected by the temperature and humidity sensor, a reference humidity is calculated based on the humidity collected by the temperature and humidity sensor, an environmental deviation is calculated based on the reference temperature and the reference humidity, and the first behavior is compensated for as a score based on the environmental deviation.

[0015] According to a second aspect of the present disclosure, a device is provided, comprising: Edge acquisition equipment, used to collect real-time environmental data, real-time behavioral data, and real-time physiological parameters; One or more processors for processing data collected by edge collection devices; A storage device is used to store one or more programs, and when the one or more programs are executed by the one or more processors, the one or more processors implement the functions of the system as described in the first aspect.

[0016] The beneficial effects of the present disclosure are: The present disclosure realizes the real-time collection and integration of physiological, behavioral and environmental data through the collaborative work of environmental collection units, behavioral collection units and wearable devices, constructs a multi-dimensional health monitoring network, and avoids the limitations of a single data source; Normal, risk, and fatigue templates are constructed to quantify the range of user behavior fluctuations. The fatigue template reflects the user's physical discomfort by prolonging the duration of the behavior and reducing the movement speed, while the risk template captures presyncope by increasing the variance of the joint angle. This effectively distinguishes between slowing movements caused by aging and abnormal states, avoiding misjudgments and improving the accuracy of anomaly detection. Based on the historical transition probability matrix, combined with the conditional probability of behavior type conversion, average time, and spatial conversion information, we determine whether user behavior conforms to spatiotemporal laws. This not only verifies the rationality of the behavior itself, but also further filters out false positives through spatiotemporal constraints, making the judgment results closer to the user's actual behavior patterns. An environmental deviation compensation mechanism is introduced to adjust behavior scores based on environmental data such as temperature and humidity. Slowing down a user's movements in high temperatures is a normal physiological reaction, and environmental deviation compensation avoids misjudging it as abnormal. Furthermore, physiological parameter anomalies caused by emotional fluctuations are corrected based on the user's emotional state, further enhancing the system's anti-interference capabilities. As an edge computing node, the data acquisition module completes data processing and storage locally on the home gateway, reducing data upload delays, enabling real-time monitoring and rapid response, and reducing the risk of sensitive health data leakage, complying with privacy protection requirements.

[0017] By setting hierarchical decision levels based on the second behavior scores, differentiated health interventions are achieved, ensuring the timeliness and rationality of responses.

[0018] It should be understood that the contents described in the Summary of the Invention section are not intended to limit the key or important features of the embodiments of the present disclosure, nor are they intended to limit the scope of the present disclosure. Other features of the present disclosure will become readily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] The above and other features, advantages and aspects of the embodiments of the present disclosure will become more apparent with reference to the following detailed description in conjunction with the accompanying drawings. The accompanying drawings are provided for a better understanding of the present disclosure and do not constitute a limitation of the present disclosure. In the accompanying drawings, the same or similar reference numerals represent the same or similar elements, among which: Figure 1 A schematic diagram of a distributed home health monitoring IoT system based on edge computing is shown. DETAILED DESCRIPTION

[0020] To make the purpose, technical solutions, and advantages of the embodiments of the present disclosure more clear, the technical solutions in the embodiments of the present disclosure will be clearly and completely described below in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present disclosure, not all of the embodiments. Based on the embodiments of the present disclosure, all other embodiments obtained by ordinary technicians in this field without making any creative efforts are within the scope of protection of the present disclosure.

[0021] In this document, the term "and / or" simply describes a relationship between related objects, indicating that three possible relationships exist. For example, "A and / or B" can represent: A exists alone, A and B exist simultaneously, or B exists alone. Furthermore, the character " / " in this document generally indicates that the related objects are in an "or" relationship.

[0022] This disclosure provides a distributed home health monitoring IoT system based on edge computing, see Figure 1 ,include: Data collection module 100: collects real-time environmental data through the environment collection unit set in the user's home, collects the user's real-time behavior data through the behavior collection unit set by the user, and collects the user's real-time physiological parameters through the wearable device worn by the user.

[0023] As an edge computing node, the data acquisition module can pre-process the collected data to improve data quality.

[0024] Specifically, the environment collection unit includes a temperature and humidity sensor; The behavior collection unit includes a door magnetic sensor, a mattress pressure sensor, a millimeter wave radar and a microphone array; The door magnetic sensor can be used to determine the user's entry and exit status, making it easier to determine the user's location. The millimeter-wave radar can be used to collect the user's movement status. The mattress pressure sensor is used to collect the user's movements in bed. The microphone array is used to collect the user's speaking voice information.

[0025] The wearable device is provided with a heart rate sensor and a skin conductance sensor.

[0026] The heart rate sensor captures the heart's electrical signals through electrodes on the skin surface to generate an electrocardiogram, which is used to diagnose heart diseases such as arrhythmia and atrial fibrillation. The skin conductance sensor quantifies the autonomic nervous system's response to emotional stimuli by detecting skin conductance activity.

[0027] As the edge computing part, the data acquisition module needs to set up a home gateway to process the collected data in real time and store the collected data for subsequent calls.

[0028] The behavior judgment module 200 is used to judge the user's real-time behavior type based on the real-time behavior data, and calculate an abnormality score based on the real-time behavior type.

[0029] Specifically, the historical behavior data collected by the behavior collection unit generates behavior templates corresponding to different behavior types; the behavior templates include normal templates, risk templates and fatigue templates; The normal template alone cannot identify whether a user's performance is slowing down due to aging, slowing down due to temporary fatigue, or even experiencing a pre-fainting condition. Therefore, three templates are constructed to quantify the range of fluctuations in user behavior. The normal template represents the user's health, the risk template represents abnormal health conditions, and the fatigue template represents a user experiencing slight discomfort.

[0030] The normal template, risk template and fatigue template all include behavior duration, maximum movement speed and joint angle variance of several types of user behaviors.

[0031] In a specific embodiment, the behavior template is constructed using the historical behavior data collected within 10 days. First, the normal feature sets of each type of behavior are clustered, and the mean and standard deviation of the duration, maximum movement speed, and joint angle variance are calculated. The mean of the three items is taken to obtain the normal template T N , see the following formula: T N = .

[0032] Among them, t is the mean duration, v is the mean maximum speed of movement, is the mean variance of the joint angles.

[0033] Fatigue Template T F In this case, the user may move slowly due to slight discomfort, so the duration is increased and the maximum speed is reduced, see the following formula: T F = .

[0034] Where k is the first expansion coefficient, is the standard deviation of the duration, is the standard deviation of the maximum speed of movement.

[0035] Risk Template T R In the case where the user faints due to physical abnormalities, the joint angle variance is increased, as shown in the following formula: T R = .

[0036] in, is the second expansion coefficient, is the standard deviation of the joint angle variance.

[0037] The corresponding real-time behavior type is determined according to the real-time behavior data collected by the behavior collection unit, and the minimum distance between the real-time behavior data and the behavior template of the corresponding real-time behavior type is calculated.

[0038] When calculating the minimum distance, the real-time behavior data needs to be matched with each template in the behavior template. The Euclidean distance can be calculated and the minimum distance can be taken as the matching benchmark. That is, if the minimum distance corresponds to the risk template at this time, it is judged that the user is in a bad state and may faint.

[0039] By constructing the above three templates and calculating the minimum distance, we can detect whether the real-time behavior type is consistent with the user's usual state and capture sudden physical abnormalities, such as whether the behavior and getting up movements are normal. This avoids misjudging the user's slow movements as abnormalities during the calculation process and reduces the false alarm rate.

[0040] Query the constructed historical transition probability matrix and calculate the reasonable value of the real-time behavior type; Based on determining whether the user behavior itself conforms to its behavioral characteristics, the present disclosure constructs a historical transition probability matrix to further analyze the user's behavior, determine whether it conforms to the laws of time and space, and distinguish whether it is the user's behavioral habits and whether the sequence of actions is normal. For example, walking out of the door late at night is abnormal behavior, thereby improving accuracy.

[0041] The historical transition probability matrix is ​​obtained in the following way: Calculate the conditional probability of converting from a real-time behavior type to the next behavior type, and calculate the average time required for the conversion. Determine the spatial conversion based on the data collected by the door magnetic sensor. The real-time behavior type, next behavior type, conditional probability, average time and space conversion are respectively used as columns, and various conversion situations are recorded in rows to obtain the historical transfer probability matrix.

[0042] The calculation formula for conditional probability P(B|A) is: P(B|A) = the number of times A to B appear in history / the total number of times A appears in history, A represents the real-time behavior type, and B represents the next behavior type.

[0043] For example, the historical transition probability matrix is ​​illustrated in the form of the following Table 1: Table 1 Example description of historical transition probability matrix

[0044] The historical transition probability matrix constructed by the query includes: After identifying the user's real-time behavior type, various conversion situations corresponding to the current behavior type consistent with the user's real-time behavior type are obtained from the historical transition probability matrix, and then waiting for the next behavior type to occur; Check whether the next behavior type meets the time and space constraints, and calculate the reasonable value of the real-time behavior type.

[0045] It's important to note that historical transition probabilities are updated every 10 days to accommodate user behavior shifts. For example, if a user adjusts their sleep schedule due to seasonal changes, the system can automatically learn the new behavior pattern, avoiding false positives caused by habit changes and improving the reliability of long-term monitoring.

[0046] The judgment of reasonable value of behavior includes two aspects: time and space. We can first judge the rationality of space. ,

[0047] Then judge the rationality of the time :

[0048] The actual consumed time refers to the time required for the completion of a real-time behavior type to the next behavior type.

[0049] The reasonable value of behavior Valid can be calculated by the following formula: Valid= · , The anomaly score is calculated based on the behavior rationality value and the minimum distance.

[0050] In a specific embodiment, the anomaly score S is calculated by the following formula: S= , in, and represents the weight and + =1, Indicates the minimum distance, Indicates the distance threshold.

[0051] Behavior modification module 300: configured to determine the emotional state based on the real-time physiological parameters and the real-time behavioral data, modify the abnormality score based on the emotional state to obtain a first behavior score; calculate the environmental deviation based on the real-time environmental data, and compensate the first behavior score based on the environmental deviation to obtain a second behavior score; Environmental factors, especially temperature and humidity, can significantly impact user behavior. For example, when the temperature is too high, users slow down their movements and minimize unnecessary activity to avoid exerting excessive force and generating excess heat. Anxiety can significantly impact heart rate and other factors, leading to misjudgments. A calm state is more likely to reflect true risk.

[0052] Determine emotional state based on real-time physiological parameters and real-time behavioral data, including: Extract features from real-time physiological parameters and real-time behavioral data, concatenate the extracted features, and output the user's emotional state after passing through a lightweight convolutional neural network; The lightweight convolutional neural network includes: a 3×1 convolutional layer, a maximum pooling layer, a 5×1 convolutional layer, a global pooling layer, and a fully connected layer; the lightweight convolutional neural network needs to be trained with a large amount of data to maximize the accuracy of the model judgment; The emotional states include negative emotions, positive emotions, and calm emotions.

[0053] The score S' of the first line can be calculated by the following formula: S'= .

[0054] Calculate the environmental deviation based on real-time environmental data and compensate the first behavior score based on the environmental deviation, including: Calculate the reference temperature based on the temperature and humidity sensor, calculate the reference humidity based on the humidity sensor, calculate the environmental deviation based on the reference temperature and reference humidity, and compensate the first behavior score based on the environmental deviation; Base temperature Calculated using the following formula:

[0055] in, Indicates the temperature collected by the temperature and humidity sensor; Baseline humidity Calculated using the following formula:

[0056] in, Indicates the humidity collected by the temperature and humidity sensor; Environmental bias Calculated using the following formula: , Second line score It can be calculated by the following formula: , Comprehensive judgment module 400: used to preset the decision level in the cloud and adopt different response methods according to the decision level corresponding to the user's second behavior score.

[0057] In a specific embodiment, the decision level indicates that different decision methods correspond to different ranges of the second behavior score. For example, when the second behavior score is between 1 and 1.5, a local reminder is made, when it is between 1.5 and 2, family members are notified, and when it is greater than 2, an ambulance and family members are notified to provide first aid.

[0058] Based on the above technical solution, the present disclosure realizes the real-time collection and integration of physiological, behavioral and environmental data through the collaborative work of the environmental collection unit, the behavioral collection unit and the wearable device, constructs a multi-dimensional health monitoring network, and avoids the limitations of a single data source; Normal, risk, and fatigue templates are constructed to quantify the range of user behavior fluctuations. The fatigue template reflects the user's physical discomfort by prolonging the duration of the behavior and reducing the movement speed, while the risk template captures presyncope by increasing the variance of the joint angle. This effectively distinguishes between slowing movements caused by aging and abnormal states, avoiding misjudgments and improving the accuracy of anomaly detection. Based on the historical transition probability matrix, combined with the conditional probability of behavior type conversion, average time, and spatial conversion information, we determine whether user behavior conforms to spatiotemporal laws. This not only verifies the rationality of the behavior itself, but also further filters out false positives through spatiotemporal constraints, making the judgment results closer to the user's actual behavior patterns. An environmental deviation compensation mechanism is introduced to adjust behavior scores based on environmental data such as temperature and humidity. Slowing down a user's movements in high temperatures is a normal physiological reaction, and environmental deviation compensation avoids misjudging it as abnormal. Furthermore, physiological parameter anomalies caused by emotional fluctuations are corrected based on the user's emotional state, further enhancing the system's anti-interference capabilities. As an edge computing node, the data acquisition module completes data processing and storage locally on the home gateway, reducing data upload delays, enabling real-time monitoring and rapid response, and reducing the risk of sensitive health data leakage, complying with privacy protection requirements.

[0059] By setting hierarchical decision levels based on the second behavior scores, differentiated health interventions are achieved, ensuring the timeliness and rationality of responses.

[0060] The present disclosure provides a device, including: an edge acquisition device for collecting real-time environmental data, real-time behavioral data, and real-time physiological parameters; one or more processors for processing the data collected by the edge acquisition device; and a storage device for storing one or more programs. When the one or more programs are executed by the one or more processors, the one or more processors implement the functions of the system.

[0061] Those skilled in the art should be aware that the embodiments described in the specification are all optional embodiments, and the actions and modules involved are not necessarily required for the present disclosure.

[0062] The program code for implementing the system of the present disclosure can be written in any combination of one or more programming languages. Such program code can be provided to a processor or controller of a general-purpose computer, a special-purpose computer, or other programmable data processing device so that when the program code is executed by the processor or controller, the functions / operations specified in the flow charts and / or block diagrams are implemented. The program code can be executed entirely on the machine, partially on the machine, as a stand-alone software package, partially on the machine and partially on a remote machine, or entirely on a remote machine or server.

[0063] In the context of the present disclosure, a machine-readable medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. More specific examples of machine-readable storage media may include an electrical connection based on one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), optical fibers, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0064] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user can provide input to the computer. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0065] The above specific embodiments do not constitute a limitation on the scope of protection of this disclosure. Those skilled in the art will appreciate that various modifications, combinations, sub-combinations, and substitutions may be made based on design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this disclosure shall be included within the scope of protection of this disclosure.

Claims

1. A distributed home health monitoring IoT system based on edge computing, comprising: Data collection module: used to collect real-time environmental data through the environmental collection unit set in the user's home, collect the user's real-time behavior data through the behavior collection unit set by the user, and collect the user's real-time physiological parameters through the wearable device worn by the user; Behavior judgment module: used to judge the user's behavior type based on real-time behavior data and calculate anomaly scores based on the behavior type; Behavior correction module: used for judging the emotional state according to the real-time physiological parameters and the real-time behavioral data, and correcting the abnormal score according to the emotional state to obtain a first behavior score; Calculate the environmental deviation based on the real-time environmental data, and compensate the first behavior score based on the environmental deviation to obtain the second behavior score; Comprehensive judgment module: used to preset decision levels in the cloud and adopt different response methods according to the decision level corresponding to the user's second behavior score.

2. The method according to claim 1, wherein The environment collection unit includes a temperature and humidity sensor; The behavior collection unit includes a door magnetic sensor, a mattress pressure sensor, a millimeter wave radar and a microphone array; The wearable device is provided with a heart rate sensor and a skin conductance sensor.

3. The method according to claim 1, wherein The method for determining the user's behavior type based on real-time physiological parameters and calculating an abnormality score based on the behavior type includes: Generate behavior templates corresponding to different behavior types through the historical behavior data collected by the behavior collection unit; the behavior templates include normal templates, risk templates and fatigue templates; Determine the corresponding real-time behavior type based on the real-time behavior data collected by the behavior collection unit, and calculate the minimum distance between the real-time behavior data and the behavior template of the corresponding real-time behavior type; Query the constructed historical transition probability matrix and calculate the reasonable value of the real-time behavior type; The anomaly score is calculated based on the behavior rationality value and the minimum distance.

4. The method according to claim 3, wherein: The normal template, risk template and fatigue template all include behavior duration, maximum movement speed and joint angle variance of several types of user behaviors.

5. The method according to claim 3, wherein The historical transition probability matrix is ​​obtained in the following way: Calculate the conditional probability of converting from a real-time behavior type to the next behavior type, and calculate the average time required for the conversion. Determine the spatial conversion based on the data collected by the door magnetic sensor. The real-time behavior type, next behavior type, conditional probability, average time and space conversion are respectively used as columns, and various conversion situations are recorded in rows to obtain the historical transfer probability matrix.

6. The method according to claim 3, wherein: Query the constructed historical transition probability matrix and calculate the reasonable behavior value of the real-time behavior type, including: After identifying the user's real-time behavior type, various conversion situations corresponding to the current behavior type consistent with the user's real-time behavior type are obtained from the historical transition probability matrix, and then waiting for the next behavior type to occur; Check whether the next behavior type meets the time and space constraints, and calculate the reasonable value of the real-time behavior type.

7. The method according to claim 6, wherein: The checking of whether the next behavior type meets the spatiotemporal constraint conditions includes: First determine the rationality of space, then determine the rationality of time, and calculate the rationality value of behavior based on the product of the rationality of time and the rationality of space.

8. The method according to claim 1, wherein Determine emotional state based on real-time physiological parameters and real-time behavioral data, including: Extract features from real-time physiological parameters and real-time behavioral data, concatenate the extracted features, and output the user's emotional state after passing through a lightweight convolutional neural network; The lightweight convolutional neural network includes: a 3×1 convolutional layer, a maximum pooling layer, a 5×1 convolutional layer, a global pooling layer and a fully connected layer; The emotional states include negative emotions, positive emotions, and calm emotions.

9. The method according to claim 1, wherein Calculate the environmental deviation based on real-time environmental data and compensate the first behavior score based on the environmental deviation, including: A reference temperature is calculated based on the temperature collected by the temperature and humidity sensor, a reference humidity is calculated based on the humidity collected by the temperature and humidity sensor, an environmental deviation is calculated based on the reference temperature and the reference humidity, and the first behavior is compensated for as a score based on the environmental deviation.

10. A device, characterized in that The device comprises: Edge acquisition equipment, used to collect real-time environmental data, real-time behavioral data, and real-time physiological parameters; One or more processors for processing data collected by edge collection devices; A storage device for storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement the functions of the system according to any one of claims 1 to 9.

Images