CN116709183A

CN116709183A - Behavior recognition system based on IOT positioning and wearing device

Info

Publication number: CN116709183A
Application number: CN202310426090.XA
Authority: CN
Inventors: 黄荣堂
Original assignee: Qiyi Platform Co ltd
Current assignee: Qiyi Platform Co ltd
Priority date: 2023-04-20
Filing date: 2023-04-20
Publication date: 2023-09-05

Abstract

The invention discloses a behavior recognition system based on an IOT positioning and wearing device, which comprises a wearing device and an IOT system, wherein a nine-axis inertial motion sensor, an altimeter, a wireless communication module and a microprocessor are built in the wearing device; the user attaches the wearing device to the chest to judge the action of the user, the nodes or labels of the IOT positioning system are installed in each region of the living space, and the packet information broadcast to the wearing device comprises: the node or the label is installed at a preset longitude and latitude height and the like at the installation position, so that the wearing device can fuse the action of a user with the information of the broadcast package, the operation is carried out in the wearing device, the accurate behavior identification of the user is obtained, the physiological and biochemical signal intelligent bracelet is further added, the physiological and biochemical signals during the behavior generation are synchronously detected, the physiological and biochemical signals are further uploaded to the server through the IOT system, and the accurate behavior analysis is carried out through the positions and the time periods of the behaviors, the physiological and biochemical signals and the habit of the user.

Description

Behavior recognition system based on IOT positioning and wearing device

Technical Field

The invention relates to the technical field of intelligent human behavior identification, in particular to a behavior identification system based on an IOT positioning and wearing device.

Background

It is known that for activities of daily living of household personnel, it is often necessary to use a camera or use microwaves or additional force sensing cushions and force sensing pedals, reed switches, or other proximity sensors to detect user interactions with various facilities in the environment. However, this requires the installation of many sensors, increasing maintenance difficulties and increasing costs. When a camera is used, privacy rights are possibly infringed, the camera is often not accepted by people, dead angles are often formed, shooting cannot be realized, and the time and resources required for processing the images are high; although the use of microwaves does not violate privacy, the microwave oven does not have dead angles, and if more than one person or other pets exist, the user cannot be judged.

In addition, in the wearing device, the wrist watch is widely used, but the wrist watch can only detect physiological signals and activity, or can simply identify human activities such as standing, sitting, lying, walking, roaming, falling, etc., but misjudgment is often not possible, such as dining, drinking water, brushing teeth, going to a toilet, bathing, watching television, making up, etc., and in addition, the judgment of the behavior needs customization or personalization, and the judgment is performed according to the location. Furthermore, if the corresponding physiological and biochemical signals may be different, for example, if the heartbeat exceeds 90 when sitting still, the heart may be very stressed, or if the heart is too high during sleep, the heart may be poor in sleep quality, or cardiovascular diseases may be caused.

TW I671740 (disclosure 202001892) is an indoor positioning system and method based on geomagnetic signals combined with computer vision. The present position and the walking track are detected by the inertial measurement unit so as to combine the computer vision coordinate and the geomagnetic signal coordinate to form a computer vision map and a geomagnetic data map. The disadvantage is that the user is required to wear the camera and to upload the cloud for a large amount of operations.

The existing indoor positioning method can roughly position indoor users, but the indoor position, such as the distance between a toilet bowl and a washbasin in a bathroom, may be less than one meter, detailed resolution is not easy, three-point positioning may be performed by using UWB, which increases installation cost, and even if the indoor positioning method is already positioned on the toilet bowl, the user cannot know whether the user faces the toilet bowl or faces away from the toilet bowl, and whether the user stands or sits down is different.

Therefore, a new invention is needed to directly let the user wear the device to judge the behavior, like using a camera to shoot an action image and using an AI to identify the behavior, the invention uses the prior living space correction setting to include the usage mode of each furniture in each region of the living space, such as facing, the action or body gesture when using, combining with the furniture function in the space, when confirming the action of the user using the specific furniture, the behavior positioning can be directly obtained, instead of the prior known technology, which requires the prior spatial positioning, then the cooperation of action detection, then the cloud conversion and the complex AI learning, so as to know the rough behavior.

In summary, it can be known that there is no currently known technology capable of directly detecting human behaviors while taking privacy into account, and thus not capable of effectively obtaining daily life information of people, and synchronously generating physiological and biochemical signals and changes thereof when carrying out the life information.

Disclosure of Invention

The invention aims to provide a behavior recognition system based on an IOT positioning and wearing device.

According to one aspect of the present invention, there is provided a behavior recognition system based on an IOT positioning and wearing device, comprising a chest wearing device and an IOT system wirelessly connected with the chest wearing device, wherein a microprocessor, a nine-axis inertial motion sensor connected with the microprocessor, an altimeter and a wireless communication module are built in the chest wearing device, a user fixedly attaches the chest wearing device to the chest to judge the actions of the user, nodes or tags of the IOT system are installed in each region of a living space, and packet information broadcasted to the chest wearing device comprises: the method comprises the steps of presetting the longitude and latitude height of an installation position, the domain name of a region where the installation position is located when a node or a tag is installed, presetting the name of each key furniture in the region where the node or the tag is located, the strength of corresponding geomagnetic fingerprints and received signals, the corresponding longitude and latitude and the azimuth of a user when using the key furniture; therefore, the chest wearing device can fuse the actions of the user with the information of the broadcast package, the microprocessor is used for directly calculating in the chest wearing device to obtain accurate behavior identification of the user, and the chest wearing device then broadcasts the acquired related information to the outside.

Further, the chest wearable device prevents falls by detecting various directions of falls and impact force of falls, or by detecting unstable standing or unstable movement.

Further, the intelligent hand ring is worn on the wrist of the user's hand, and synchronously detects physiological signals and biochemical signals between the physiological signals and the biochemical signals, and broadcasts the physiological signals and the biochemical signals to the outside, wherein the physiological signals are selected from at least one of heartbeat, electrocardiogram, body surface impedance, body temperature, blood oxygen and blood pressure, and the biochemical signals are selected from at least one of blood glucose concentration, lactic acid concentration, alcohol concentration, corticoid concentration and drug concentration.

In some embodiments, the chest wearing device further comprises at least one of a camera or a recording device, the camera or the recording device can be triggered to shoot or record by the action judgment or the position judgment of the wearing device, or the user can shoot or record by triggering a button of the chest wearing device by using a Bluetooth, or the camera or the recording device is installed inside the chest wearing device by using the external world, or a wireless camera or a wireless recording device is additionally arranged to be worn on the user.

Specifically, the chest wearing device further comprises at least one of a camera or a recording device, the camera or the recording device is triggered and started by the intelligent bracelet, the distance between the intelligent bracelet and the chest wearing device is close because a user holds an article close to the mouth, the receiving signal intensity between the intelligent bracelet and the chest wearing device is relatively enhanced, and the camera or the recording device is triggered and shot or recorded by a secondary trigger.

Further, the IOT system further comprises a router and a server, the router receives broadcast information packets of the chest wearable device and the smart band and uploads the broadcast information packets to the server, the server can know the position and time period of occurrence of related actions of the user and habits of the user, and the server analyzes and records the related actions, further analyzes the correlation between physiological signals and biochemical signal changes during the occurrence of the related actions, analyzes the stability and variability of life activities of the user, analyzes the variability of the related actions, or analyzes the variability of specific activities.

Specifically, the procedure of presetting the packet information broadcasted to the chest wearable device is as follows:

s1, an APP is utilized to assist a user to establish an indoor plane modeling diagram of a living space, wherein the plane modeling diagram comprises positions of furniture in each area, and a socket position of a node to be installed or a position of a wall to which a label powered by a battery is to be attached;

S2, installing the nodes to socket positions of each area according to the indoor plane modeling diagram; or attaching a label to the indoor wall of each area;

s3, a user wears the chest wearing device, geomagnetic fingerprints, received signal strength and longitude and latitude of each piece of furniture are set for each piece of furniture in each area, and the direction of the user when the user uses the corresponding piece of furniture is set.

Further, a pedometer is further arranged in the chest wearing device, the nine-axis inertial motion sensor and the altimeter calculate dead reckoning through furniture and give a reliable position, when a user leaves the position, the position is a starting point, the pedometer and the microprocessor are used for judging the facing direction of the user, the read values are put into the dead reckoning to calculate the next position, and geomagnetic fingerprints of the position, the strength of received signals, the longitude and latitude and the azimuth of the user are combined, so that the chest wearing device obtains the next furniture closest to the user through fusion calculation, and accurate behavior identification specific data of the user are obtained through actions of the furniture.

Further, the detailed steps of presetting the package information broadcasted to the chest wearable device are as follows:

S1, a user wears a device on the chest and carries a handheld mobile device provided with a corresponding APP to go to a certain room;

s2, standing in front of a piece of furniture or using the piece of furniture, staying still, starting pressing setting in an APP interface of the handheld mobile device, informing the chest wearing device by the APP, broadcasting the setting mode to the APP by the chest wearing device, wherein the broadcasting content of the chest wearing device comprises the received signal intensity of a positioning broadcasting packet sent by a node or a label at the time, the azimuth of the chest wearing device and the geomagnetic intensity of the chest wearing device, accumulating at least 20 pens after the APP receives the broadcasting content of the chest wearing device, averaging the received signal intensity, the azimuth and the geomagnetic intensity, adding the longitude and latitude of the furniture position, and setting the fingerprint information of the furniture position in the room;

s3, after 5 seconds, the APP informs the user of finishing correction, and the user goes to the next furniture;

s4, repeating the step S2 and the step S3 until all furniture is corrected, and obtaining fingerprint information;

s5, connecting the APP with the node or the label, and writing the fingerprint in the step S4;

s6, going to another room, repeating the steps S2 to S5 until all rooms are completed.

Specifically, the camera is a dining table camera, the camera is erected in the range of 80-150cm above the dining table, the dining table camera is a 3D camera, the shooting range of the camera is only a dining table area, as long as a user stays near the dining table, no matter sitting, the chest wearing device transmits dining table positioning information to the cloud end, the cloud end push requests the dining table camera to shoot a photo, the cloud end is uploaded, and the behavior of the dining table is matched with the content of the photo to judge the activity type and the diet content of the user.

In some embodiments, the chest wearing device and the smart band can be worn respectively, wherein the chest wearing device is fixed in the chest in a safe form, and then the necklace is attached, so that the device can be worn for a long time, habit can be easily developed, the device is not easy to place randomly, the wearing can not be forgotten, the device is easy to accept, and particularly if the device can be used for ensuring safe detection of falling, preventing falling and having an emergency help button for asking for help. Thus, daily life and body posture can be detected by the wearing device fixed on the chest. The smart band adopts bluetooth broadcasting, the security symbol has transmission capability of more than 5.0 bluetooth, and can use the encoded PHY to transmit long packets, the security symbol can receive the broadcasting packets of the smart band, and then integrate the body gesture actions obtained by the security symbol and the positioning fingerprints into one packet, and upload the packet to the router or gateway and then upload the packet to the server or cloud.

Further, the security symbol can also detect the posture of the body and can also be used for interacting with the smart band to detect whether the smoke is being drawn by the hander. The method is that the intelligent bracelet is worn on a conventional hand, physiological signals can be broadcast to the safe symbol after being fixed for a period of time, for example, 1-2 seconds, if smoke is not drawn, the intelligent bracelet is far away from the safe symbol, and the intensity of Bluetooth communication receiving signals between the intelligent bracelet and the safe symbol is weak; however, when smoking, the intelligent bracelet is close to the safe symbol, and the intensity of Bluetooth communication receiving signals between the intelligent bracelet and the safe symbol is relatively enhanced. The period of the negligence of the broadcast packet of the intelligent bracelet received through the safety symbol is close to the period of the smoking approaching the mouth and leaving the mouth, so that the user can be judged to be smoking. The smoking behavior, time and times of occurrence are recorded every day, so that the user can know the smoking addiction, and the physiological data of the user can be synchronously compared before smoking and during smoking and after smoking, or the reason of smoking can be found, and a smoking stopping strategy is provided.

Further, the camera can be triggered to take a picture by the action judgment or the position judgment of the security symbol, or the user can trigger to take a picture by the button of the wearing device, or take a picture by the external world through Bluetooth, or take a picture by the dominant hand wearing the smart bracelet, because the object is held close to the mouth, the smart bracelet is close to the security symbol, the strength of the Bluetooth communication receiving signal between the smart bracelet and the security symbol is relatively enhanced, and the camera is triggered to take a picture. One of the important applications is the recording of eating, taking medicine, drinking water, drinking beverages, smoking cigarettes, etc., because the behavior and content of the diet are vital to health, and the user who is engaged in the diet anytime anywhere can effectively record the behavior of the diet and the diet content thereof every time by a camera hung in the chest.

Furthermore, the wearing device can be added with at least one of a Bluetooth small camera and a sound recording device besides the safety symbol hung on the chest, the Bluetooth small camera can be a video pen or a sound recording pen, or is built in glasses, the Bluetooth camera can be triggered to take a picture by the action judgment or the position judgment of the safety symbol, or is triggered to take a picture by a user on the safety symbol button, or is triggered to take a picture by a handy hand wearing the smart bracelet because a holding object is close to the mouth, the distance between the smart bracelet and the safety symbol is close, the strength of Bluetooth communication receiving signals between the smart bracelet and the safety symbol is relatively enhanced, and the Bluetooth of the safety symbol triggers the camera to take a picture. Or a proximity sensor such as infrared ranging is additionally arranged in the wearing device, and the proximity sensor triggers the camera to take a picture because the object is held close to the mouth, and therefore, the object is also close to the wearing device in front of the chest.

From the past literature, it is known that the wearing device is to be able to accurately determine the movements of the user, such as standing, sitting, lying, walking, roaming, falling, running, etc., where the best wearing position is not the chest, and needs to be fixed to the chest without being able to walk or shake at will. However, the chest strap is often used for fixing, but the method is difficult to wear for 24 hours, and if the user's behavior is to be detected and monitored completely at any time, particularly if the user is to prevent and detect falling, no empty window period is needed. There are also drawbacks in the conventional art using the hanging wearing device, in that the hanging wearing device has a walking position and its posture cannot be synchronized with the posture of the upper torso of the user, so that the body's movement and posture cannot be reflected, for example, the standing and sitting may not be effectively distinguished, or the upper torso is bent forward, and the angle of the three axes of the hanging wearing device is not different from that of the upper torso, so that it is difficult to recognize whether the wearer is bent. When running, the wearing device hung at possibly can shake randomly, and the running steps and the like cannot be accurately judged.

The invention provides a method for effectively fixing the skin of the chest of a wearer, which is characterized in that the requirements of wearing safety marks on the body of the user are that the safety marks are fixed on the body, preferably the chest, in order to ensure that the temperature is high in summer, the clothes are few, and the safety marks can be worn for 24 hours when the user is not wearing clothes or only wearing swimming trunks, and the method comprises the following steps:

the bionic self-adhesive-free structure, such as gecko feet, octopus sucking discs and the like, is used as a back adhesive film of a safe sign, can effectively grasp the chest skin of a user, but retains sufficient air permeability, is easy to take down, cannot cause skin injury, and can be taken down at any time and then adhered back to the chest skin. The bionic glue-free self-adhesive structure at least maintains the viscosity for three to seven days, and can be worn continuously even when in bath, so that a user can be protected, and the bionic glue-free self-adhesive structure can be used for detecting falling in a bathroom which is easy to fall and even detecting unstable standing.

And step two, further, the neck hanging wire is connected with the safety sign, so that the safety sign can be naturally hung when being taken down or replaced, and is uniformly fixed at the position in front of the chest, and the bionic non-adhesive self-adhesive structure in the step one can be avoided, and the safety sign can fall off the skin due to external force impact. If the clothes are worn more, the clothes are hung on the neck and can not slide, and the safe symbol can be fixed with the clothes by using the velcro.

The method of fixing the security symbol according to the present invention can provide a number of advantages, such as:

1. the judging reference value of various actions is unchanged, because the safety symbol is fixedly attached to the chest of the user, the problems of walking and random shaking of the wearing device which is simply hung do not occur, the gesture obtained by the measurement values of the IMU and the altimeter on the safety symbol is completely synchronous with the gesture of the upper half trunk of the user, the actions and the gesture of the body can be specifically reflected, for example, the standing and sitting can be effectively distinguished, or the lying on the bed and the lying on the ground can be identified by adding the altimeter. When running, the wearing device can not shake at will, and the running steps, postures and the like can be accurately judged

2. The user can be informed whether to be unstable or restless when standing, whether to shake the body left and right when walking, and whether to fall risk, because the gesture obtained by the microprocessor on the safety sign and the measurement value of the altimeter is completely synchronous with the gesture of the upper half body of the user, the gesture can be compared with the reference value of various actions, and if the deviation value is overlarge and exceeds the threshold value, a warning can be sent.

In some embodiments, the security symbol is also compatible with the function of a smart bracelet, is integrally formed, can be attached to the skin in front of the chest, and can simultaneously complete human activity recognition, sleep depth recognition, heartbeat or electrocardiogram, respiration, skin impedance and body temperature. Further, the quality of sleep and the presumed emotion can be evaluated.

The peace sign and smart wristband can be used to obtain measurements, such as respiration, blood oxygen or electrocardiogram, to estimate sleep quality and sleep apnea.

The basic functions of security symbol and smart band integration allow the user to include 1: prevent and detect falls; 2. healthy: normal work and rest, sufficient sleep, normal medicine taking, clever exercise, water drinking, balanced diet, toilet, defecation, normal physiological signals and bath; 3. mental health: the emotion is good, and the body surface impedance is normal when people eat three meals, heart beat and blood pressure.

The function of the security symbol may send out broadcast packets containing position, motion, and body posture (steady or shaking) and body temperature (optionally provided) at regular intervals (e.g., 30 seconds).

The IOT system comprises a tag (Beacon), a router, a Gateway (Gateway) and a server; the beacon is composed of a Bluetooth 5.0 module, at least one of each room at home needs to be installed, the beacon is powered by a battery, and the beacon can be attached to the ceiling or the wall of each room along with the application of the beacon; the packets broadcast by the beacon are shown in table 1; the security symbol and the intelligent bracelet adopt Bluetooth 5.0 and directly upload a router/gateway (router/gateway) with a Bluetooth 5.0 module, so that the system is simplified, the cost is reduced, for example, 10 rooms or areas are reduced, 10 beacons are needed, the price of the 10 beacons is about 2000 yuan, and the price of the router/gateway is about 2000 yuan; the beacon uses the battery with larger capacity, broadcasts every second for 0.1 second, can change once a half year, even change once a year, can inform the router/gateway when not having the electricity, and then inform the server to relay the replacement battery of the user. The security symbol uses a multi-role bluetooth function to receive beacon broadcast packets and conduct behavior recognition from IMU measurements. In some examples, more than three beacons may be attached to each room or region, providing three-point positioning. In some examples, only the version of more than 5.0 bluetooth may be used as the security symbol, while the version of 4.0 bluetooth may be used as the smart band, the smart band is broadcasted to the security symbol, and the security symbol converts the measured value of the smart band and the action and behavior information measured by the security symbol into a long packet of 5.0 bluetooth, most often up to 245bytes, and the long packet is directly transmitted to the router/gateway and the cloud server.

Dining table cameras use 3D cameras, such as Intel RealSense-D430, to purposely capture activities on the dining table, i.e. according to daily schedule, GOOGLE calendar starts, but for random eating activities, it is not possible to track. The random diet is touched through positioning, so long as a user stays at the dining table area, no matter sitting, the safe sign transmits dining table positioning information to the cloud end, the cloud end push requests the camera to take a picture, and the picture is uploaded to the cloud end, and the shooting range is only the dining table area, so privacy is not interfered. The behavior in a restaurant may also require a table camera to determine its type, such as eating, taking medicine, drinking water, eating snacks, etc.

In other embodiments, a smart speaker may be further included, with a master bedroom with a guest/restaurant, to actively alert the user, or to passively answer the user's query.

The intelligent sound box can be a mobile intelligent sound box, such as a mobile robot of Zenbo, temi, amazon Astro, a robot dog and the like of Hua Shuo, and the mobile robot comprises a camera besides the sound box, can identify the environment and a user, can follow the user, can receive the command of a cloud server when the physiological or biochemical signals of the user are abnormal, and can go to the position of the user, monitor nearby and actively remind the user, or answer the inquiry of the user passively, but need to pay attention to privacy issues possibly generated by the camera. It should be noted that the machine dog can climb stairs and can stay outdoors, so that the signals obtained by the user's safety symbol and the smart bracelet can be directly transmitted to the machine dog for transfer to the cloud or directly operated by the processor on the machine dog, and thus, the use situation of the invention can be expanded to outdoor activities.

The first function of the server or cloud is to judge the behavior; the cloud may be required to collect data from the security symbol (or smart band) after a certain period of time (usually the user leaves the location and ends the action), so as to determine correctly, for example, go to a toilet, possibly a large size or a small size, too long in a large size, constipation, and multiple times in a large size, and possibly pull the abdomen. The small number typically lasts no more than 3 minutes. After the toilet is finished, the user should turn around to wash hands on the wash station.

The second function of the server or the cloud is to judge the relevance of the behavior and the physiological signal; reference is made to customized physiological signal monitoring and alerting.

The third function of the server or the cloud is to conduct reminding operation through the sound box or the APP according to daily preset regular activity work and rest, and confirm whether the preset activity is completed or not through feedback of the safety symbol, if not, continuously remind until the preset activity is completed, and update a real work and rest table. For example, the taking of medicines, which the genus has to do, must be completed.

The fourth function of the server or the cloud is to provide a preset activity and work table every day, complete the real activity and work table every day in the early morning, and fill physiological data and action gestures when the activities or actions in the work table are performed, namely quality factors of the corresponding actions.

Drawings

Fig. 1 is a flowchart of an IOT positioning and wearable device-based behavior recognition system according to an embodiment of the present invention.

Fig. 2 is a schematic diagram of an APP setting field of the IOT positioning and wearable device-based behavior recognition system shown in fig. 1.

Fig. 3 is a schematic diagram of a positioning fingerprint of the APP set indoor room of fig. 2 for a number of important positioning points.

FIG. 4 is a schematic diagram of the relationship between the fingerprint and behavior of the APP set up indoor important furniture in FIG. 2;

fig. 5 is a schematic circuit diagram of a wearable device of the IOT positioning and wearable device-based behavior recognition system shown in fig. 1.

Fig. 6 is a state machine diagram of the IOT positioning and wearable device-based behavior recognition system shown in fig. 1.

Fig. 7 is a partial behavior action recognition flow chart of the IOT positioning and wearable device-based behavior recognition system shown in fig. 1.

Fig. 8 is a schematic diagram of detection results of a backward fall while standing of the IOT positioning and wearable apparatus-based behavior recognition system shown in fig. 1.

Fig. 9 is a schematic diagram of detection results of left fall detection during standing of the IOT positioning and wearable apparatus-based behavior recognition system shown in fig. 1.

Fig. 10 is a graph showing changes in physiological data of a subject over an hour at a time point of the day.

Fig. 11 is a schematic diagram of the posture distribution of the subject at a certain time point of the day for one hour.

Fig. 12 is a graph comparing experimental results of TUG experiments performed under the same conditions for young and old.

Fig. 13 is a schematic diagram of setting 8 daily home environmental behavior events and measuring RSSI signal strength.

Fig. 14 is a behavior trace diagram of 8 set daily home environmental behavior events.

In the drawings, reference numerals:

16 battery 17 emergency button

18 circuit board

181 nine-axis inertial sensor 182 air pressure height sensor

183 antenna 184 body temperature sensor

Router for 21 node 23 Bluetooth to convert WIFI

61 toilet 63 user

67 wash-stand

Detailed Description

The invention is described in further detail below with reference to the accompanying drawings. Embodiments of the present invention are described in detail below, examples of which are illustrated in the accompanying drawings, wherein like or similar reference numerals refer to like or similar elements or elements having like or similar functions throughout. The embodiments described below by referring to the drawings are illustrative and intended to explain the present invention and should not be construed as limiting the invention.

Fig. 1 to 7 schematically illustrate a behavior recognition system based on IOT positioning and wearing apparatuses according to an embodiment of the present invention.

As shown, the system includes a chest wearable device and an IOT system wirelessly connected with the chest wearable device.

The user 63 attaches the chest wearable device to the chest to determine that the node 21 or tag of the action IOT system of the user 63 is installed in each region of the living space, and the packet information broadcast to the chest wearable device includes: the longitude and latitude height of the installation position, the domain name of the area where the installation position is located, the name of each key furniture in the area where the installation position is located, the corresponding geomagnetic fingerprint and received signal strength, the corresponding longitude and latitude and the azimuth of the key furniture used by a user are preset when the node 21 or the tag is installed; therefore, the chest wearing device can fuse the actions of the user 63 with the information of the broadcast packet, and directly calculate in the chest wearing device through the microprocessor to obtain accurate behavior identification of the user 63, and then the chest wearing device broadcasts the acquired related information to the outside.

The chest wearing device prevents falls by detecting various directions of falls and impact force of falls, or by detecting unstable standing or unstable movement.

In this embodiment, the chest wearable device is a peace sign.

In this embodiment, the system further comprises a smart band worn on the wrist of the dominant hand of the user 63, and synchronously detecting physiological signals and biochemical signals between the physiological signals and the biochemical signals, wherein the physiological signals are selected from at least one of heart beat, electrocardiogram, body surface impedance, body temperature, blood oxygen or blood pressure, and the biochemical signals are selected from at least one of blood glucose concentration, lactic acid concentration, corticosteroid concentration or drug concentration.

In other embodiments, the chest wearable device further includes at least one of a camera or a recording device, where the camera or the recording device may be triggered to take a photograph or record by the action determination or the position determination of the wearable device, or the user 63 may take a photograph or record by triggering the emergency button of the chest wearable device, or the camera or the recording device may be installed inside the chest wearable device by external triggering via bluetooth, or a wireless camera or a wireless recording device may be additionally provided and worn on the user 63.

The camera or the recording device can be triggered and started by the intelligent bracelet, and the hander wearing the intelligent bracelet is close to the chest wearing device because the object is held close to the mouth, so that the distance between the intelligent bracelet and the chest wearing device is close, the strength of a receiving signal between the intelligent bracelet and the chest wearing device is relatively enhanced, and the shooting or recording is triggered by times.

Further, the IOT system further comprises a router and a server, wherein the router receives broadcast information packets of the chest wearable device and the smart band and uploads the broadcast information packets to the server, the server can know the position and time period of the related behavior of the user 63 and the habit of the user, and the server analyzes and records the related behavior, further analyzes the correlation between the physiological signal and the biochemical signal change during the related behavior, analyzes the stability and variability of the life of the user, analyzes the variability of the related behavior, or analyzes the variability of specific activities.

Specifically, the procedure of presetting packet information broadcast to the security symbol is as follows:

s1, an APP is utilized to assist a user to establish an indoor plane modeling diagram of a living space, wherein the plane modeling diagram comprises the position of each furniture in each area and the socket position of a node 21 to be installed or the position of a wall to which a label powered by a battery is to be attached;

s2, installing a node 21 to the socket position of each area according to the indoor plane modeling diagram; or attaching a label to the indoor wall of each area;

s3, the user 63 wears the security symbol, geomagnetic fingerprints, received signal strength and longitude and latitude of each piece of furniture are set for each piece of furniture in each area, and the orientation of the user when the user uses the corresponding piece of furniture is set.

Further, a pedometer is further arranged in the safety symbol, the nine-axis inertial motion sensor 181 and the barometric pressure sensor 182 calculate dead reckoning through furniture and give a reliable position, when the user 63 leaves the position, the position is a starting point, the pedometer and the microprocessor are used for judging the face orientation of the user 43, the read values are put into the dead reckoning to calculate the next position, and meanwhile, the geomagnetic fingerprint of the position, the intensity of received signals, the longitude and latitude and the azimuth of the user are combined, so that the safety symbol obtains the next furniture closest to the user 63 through fusion calculation, and accurate behavior identification specific data of the user 63 are obtained through actions of the furniture.

Specifically, the camera can be a dining table camera and is arranged in the range of 80-150cm above the dining table. The dining table camera is a 3D camera, the shooting range of the dining table camera is only a dining table area, as long as a user 63 stays near the dining table, no matter sitting, the safety sign transmits dining table positioning information to the cloud end, the cloud end push requests the dining table camera to shoot a photo, the photo is uploaded to the cloud end, and the behavior of the dining table is matched with the content of the photo to judge the activity type of the user 63.

The operation sequence of the present invention, referring to FIG. 2, includes the steps of setting up a new field; step two, users wear the security symbol and the intelligent bracelet, and data are collected in daily activities; thirdly, the server or the cloud performs analysis and operation on data collected by the security symbol and the intelligent bracelet, even the environmental sensor, accurately identifies various behaviors, and classifies good habits and bad habits of the behaviors by the history record; and step four, selectively implementing intervention of behavior change, and repeating the step two and the step three to confirm whether the behavior is improved or not.

The three steps are described in detail below.

Step one, setting installation of a new field:

referring to FIG. 2, a preferred approach may use [ installation settings APP ], which provides the following functions:

The user 63 is assisted in building an indoor plan modeling map of the indoor area, including the location of each furniture in each room, and the socket location. The socket location may be used to install a bluetooth communication Node (Node) 21. If battery powered bluetooth tags (beacons) are used, they can be attached to indoor walls.

With the indoor plane modeling diagram, a field user installs a Node (Node) 21 to the outlet location of each room. Or a label (beacon) is attached to the indoor wall of each room. And a router/Gateway (bwrlutter/Gateway) 23 for bluetooth to WIFI conversion is installed in a socket or the like with a mains supply, a preferred example of an installation location is a central location of a home area or above a dining table, so that a camera above the dining table is installed at the same time. The router/gateway for bluetooth to WIFI conversion may be built by a module above bluetooth 5.0 in combination with a Raspberry PI module (e.g., raspberry PI 4).

The user 63 wears the security symbol, and sets his geomagnetic fingerprint, received signal strength, longitude and latitude, and user orientation to each furniture in each room, thereby completing the contents of table 1.

Table 1 home node packet information table

The detailed steps of the packet information broadcast to the security symbol are as follows:

S1, a user 63 wears the mobile device as a security symbol and carries the handheld mobile device with the corresponding APP, and the mobile device goes to a certain room;

s2, standing in front of a piece of furniture or using the piece of furniture, staying still, starting to press setting in an APP interface of a handheld mobile device, informing a security symbol by the APP, broadcasting a setting mode to the APP by the security symbol, wherein the content of the security symbol broadcasting comprises the received signal intensity of a positioning broadcasting packet sent by a node 21 or a label at the moment, the azimuth of the security symbol and the geomagnetic intensity of the position where the security symbol is located, accumulating at least 20 pens after the APP receives the broadcasting content of the security symbol, averaging the received signal intensity, the azimuth and the geomagnetic intensity, adding the longitude and latitude of the furniture position, and setting the fingerprint information of the furniture position in the room;

s3, after 5 seconds, the APP informs the user of finishing correction, and the user 63 goes to the next furniture;

The field correction requires direct geomagnetic numerical detection by a person wearing the security symbol, and preferably the posture is adapted to the habit of using the furniture, for example, the toilet 61 should be standing facing the toilet 61 or sitting away from the toilet 61. For example, the use of the wash station 67 should be such that it stands facing the wash station 67 and then brushes, washes the face, washes the hands, brushes the beards, etc. For example, the use of a bathtub and a shower faucet can be used in multiple directions.

Step two, the user wears the security symbol and the intelligent bracelet, and data are collected in daily activities:

the user wears the security symbol to enter a certain room, receives the home node information broadcast packet of the table 1, obtains geomagnetic information from the position, obtains the room name and the furniture name of the table 3 through fingerprint comparison with the table 1, and sends the broadcast packet of the table 2 to the router 23 (bwrluter) directly to upload to the cloud after the tables 4 and 3 are completed; or in a larger area, the router is reached through the node 21 or the tag and then through the bluetooth gate, and the router is uploaded to the cloud. The smart band may be further enhanced with measuring biochemical signals, which may be referred to by the inventor, taiwan patent I730503 "physiological and biochemical monitoring device", thus generating physiological data and biochemical data embodiments as shown in table 5, wherein the physiological data may generate a message in 1-5 seconds, and the biochemical signals are typically generated in 1-10 minutes, preferably every 5 minutes. In certain embodiments, the physiological signal is selected from the group consisting of heart beat, electrocardiogram, HRV, body surface impedance, body temperature, blood oxygen, blood pressure, and the biochemical signal is selected from the group consisting of blood glucose concentration, lactate concentration, cortin concentration, drug concentration. In some embodiments, the biochemical signals can also be measured using a micro-spectrometer and integrated into the bracelet with the physiological signal sensor for real-time continuous monitoring of the physiological and biochemical signals.

Table 2 principal humanoid security symbol broadcast packets

TABLE 3 thing-to-location related behavioral interpretation of security symbols

/>

Table 4 sensed information interpretation of security symbols

/>

TABLE 5 physiological data and Biochemical data examples

Since the fingerprint data of table 1 is used to determine a certain location in a certain room, in some cases, there is still no high resolution of location, for example, at least one meter. Therefore, the positioning function of the security symbol itself can be used, that is, by using a state machine, the action of the position a is definitely defined, for example, in a toilet, the action of sitting down can be estimated, the position a is the position of the toilet, then the user stands up, walks around, and goes to the position B, although the geomagnetic fingerprint, the received signal strength and the azimuth of the position B are similar to those of the position a, the processor can judge that the user is at the position B due to judgment of the state machine. That is, when there are multiple locations in room a, it is possible that the fingerprints of two locations are similar, so that the processor can estimate the next location by determining the location (latitude and longitude) of the previous state, and the transformation process is determined by the change of the pose of the security symbol. I.e. to calculate a new position (longitude and latitude) via the walking trajectory. Thus, the fingerprint of the position (longitude and latitude) can be added to accurately perform positioning, particularly behavioral positioning.

Further, referring to fig. 3, which is a schematic diagram of a positioning fingerprint of an important positioning point of an indoor room, the wearable device (security symbol) of the present invention can calculate a track of relative movement by using an accelerometer, a gyroscope and a compass, so that the track can be calculated by using a pedometer plus the accelerometer, the gyroscope and the compass as long as a static starting point is used as a reference, and the track can be represented by longitude and latitude.

In other words, in this architecture, because the principle of using dead reckoning (Pedestrian Dead Reckoning, PDR) is to calculate the next position using the sensor to calculate the number of steps and the dead reckoning angle at a known position, or to calculate the walking distance using the pedometer and the steps, in this embodiment, many fixed furniture such as toilets, washbasins, bathtubs, gas stoves, couch etc. are used, which are not normally moved, so that the furniture can give a reliable position, which is the starting point when the user leaves the position, and the pedometer and the processing module determine the orientation, and put the read values into the PDR to calculate the next position.

Referring to FIG. 4, a schematic diagram of the relationship between the fingerprint and behavior of the important furniture in the indoor room according to the present invention is shown. Node 21 actively broadcasts the data packet to the user's security symbol.

Taking the toilet 61 as an example, if the user 63 is a male, the user will stay on the toilet 61 and will show a small-size behavior. If a male user 63 is facing away from the toilet 61 and has a sitting down action, and stays a little longer, it may be a large-sized action. If the user 63 is facing the toilet table 67, the user will act to wash his face, wash his hands, or brush his teeth.

The security Fu Najian of the invention can acquire BLE receiving signal intensity and geomagnetic fingerprint from the communication node 21 or the tag with respect to the shaft IMU and the altimeter, and calculate PDR by adding the processing module of the security symbol, so that sufficient positioning accuracy can be achieved, the required communication node or tag density can be said to be the most simplified scheme in practice, only one communication node or tag in one room is needed, and the Bluetooth is not required to provide the function of three-point positioning.

In the embodiment of the invention, a signal line comparison method can be adopted, a characteristic database is established in an off-line stage, and basically, n reference points, preferably 4-5 reference points, exist in each area or room. The feature data of one reference point includes position (longitude and latitude), x (x-axis geomagnetic component), y (y-axis geomagnetic component), z (z-axis geomagnetic component) and F (integrated geomagnetic intensity), received signal strength, and azimuth as shown in table 1. When the current feature data is received in the online stage, the data is compared with all features of the feature database to estimate the current position of the user, and the steps are as follows:

In the first step, it is assumed that n reference points are provided, and the corresponding sampling point feature data includes geomagnetic intensity Fi, bluetooth communication intensity received signal intensity i, azimuth i, longitude and latitude (Pxi, pyi), i=1. The average intensity of each feature component is calculated first, and then the standard deviation of each feature component is calculated.

The second step, for those that can produce larger standard deviations, indicates that their identification capability is better. Therefore, a larger weight is given, the weight W is set as the sum of the standard deviation of each characteristic, and the weight of each axis is changed to the ratio of W.

And thirdly, when receiving the geomagnetic components of each axis of the current position of the user, comparing the geomagnetic components with the characteristic data of each axis in the off-line stage. And calculating the minimum difference value between the calculated position and the characteristic data to estimate the position P of the corresponding reference point, and estimating the current position of the user.

Fourth, the portable device security sensor of the present invention can receive 5 to 10 feature data per second. Therefore, 5 to 10 position results can be calculated from the method of the third step in one second, the position results are used for calculating the optimal position by using the KNN method, and the behaviors of the user are calculated from the furniture corresponding to the position and the actions of the user, as shown in the table 3 and the table 4.

The safety symbol of the present invention has a circuit board 18 in which a nine-axis inertial sensor (BMX 055) 181, an atmospheric pressure sensor (BMP 280) 182, a body temperature sensor 184, and other sensors, and a processing module and a wireless communication module are built.

The circuit board 18 is powered by a battery 16 and has an antenna 183.

The chest wearable device is also provided with an alarm emergency button 17.

The processing module (CPU) includes a bluetooth module.

The nine shafts comprise a triaxial accelerometer, a triaxial gyroscope and a triaxial magnetometer, are applied to gesture recognition and limb rotation, can respectively measure acceleration, angular velocity during rotation and geomagnetic direction, further calculate the rotation angle and moving distance of the device, and can learn whether the current behavior is sitting, standing or walking and other states by the wearable device itself for positioning.

The method comprises the steps of linearly fusing the calculated postures of the accelerometer and the magnetometer with the calculated postures of the gyroscope through a Madgwick algorithm (gradient descent algorithm), finally obtaining the optimal postures, obtaining Roll, pitch, yaw three-axis rotation angles with high posture accuracy in the mode, calculating resultant force G values and Pitch values of the three-axis acceleration square root numbers to judge sitting and standing behaviors, detecting relative height changes by combining the barometer, and assisting in distinguishing the postures and recognizing the behaviors.

Referring to fig. 6, switching of behavior states is performed using a Finite State Machine (FSM), and corresponding events are performed according to the current behavior state. Finite state machines, also known as finite state automata, represent mathematical computational models of finite states and behaviors such as transitions and actions between these states, and have wide application in the computer field. FSM is a high-efficiency compiling method in logic unit, in the course of server compiling, the server can make correspondent logic treatment according to different states or message types so as to make the program logic clear and easy to understand. FSM is divided into Moore state machine and Mealy state machine, the former refers to finite state automaton whose output is determined by current state only, namely, the output can be regarded as the function of current state; the latter is a finite state automaton that generates output based on its current state and input, and the output signal is related to not only the current state but also all input signals, so that the output can be regarded as a function of the current state and all input signals, and since the behavior of a person is changed in many ways, some behaviors are unpredictable, as shown in fig. 6, the states of standing, sitting, walking, falling and the like are first divided, when the current behavior state is standing, the next behavior prediction may be falling or sitting, and the like, but falling belongs to an emergency, and falling may occur in one second in any behavior mode, or some abrupt conditions, so that the safety symbol program is written in terms of the concept of a Mealy state machine, and further how to switch the behavior modes is judged. And comparing the input state with the current state, executing the corresponding function action as long as the correlation of the behavior state is met, and jumping to the next state. The method comprises the steps of firstly setting the wearable device to be in an initial state when the current state is not known to be sitting, standing or walking when the wearable device is started, and determining the current state according to the behavior state executed for the first time. For example: wearing a wearing device when a person stands, if the next behavior state is standing or sitting, executing a standing function, and jumping the initial state of the current state to the standing state; if the current state is up, the input state is sitting, and the state machine is out of logic in comparison, the corresponding sitting function is not executed, and the user cannot jump to the sitting state, so that the behavior restriction is performed in the mode, and the probability of misjudgment is reduced.

In some embodiments, a pneumatic altimeter may be further added to the sensor of the security symbol in addition to the nine-axis IMU. And the sensor fusion technology is used for judging the accuracy of various behaviors. Referring to FIG. 7, a partial behavioral action recognition flow chart according to an embodiment of the invention.

Fall detection:

1. altimeter (average barometric pressure value) change: from 120-150cm (standing), to 15-25cm (falling), at least over 80cm, and correction according to the height of the wearer is required. Judging the IMU;

2. the Picth angle is changed from 0 degree to 80-90 degrees;

3. ax is monotonically changed from-1 g to 0.

Fall prediction and prevention:

1. pitch > + -5 gradient/step occurs in 70% in 10 consecutive steps;

2. roll > + -5 gap/step occurs 70% in 10 consecutive steps.

Walking detection:

1. the altimeter is larger than 120-150cm (corrected according to the height of the wearer);

2. the Az of the IMU has a periodic variation in height.

Step counting function:

2. the Az of the IMU has a periodic variation in height. The steps are counted once per cycle.

Standing detection:

2. The IMU signal or detecting the stop from walking to sitting to standing.

Sitting detection:

1. 80-90cm (corrected according to the height of the wearer);

2. an IMU signal from standing to sitting is detected.

Detecting when getting up:

1. 80-90cm (corrected according to the height of the wearer);

2. detecting an IMU signal from lying to sitting;

3. ax monotonously changes from 0 to-1 g.

Bedridden detection:

1. altimeter 40-50cm (corrected according to the height of the wearer);

2. detecting an IMU signal from sitting to lying;

3. ax is monotonically changed from-1 g to 0.

Referring to fig. 8, an embodiment of backward fall detection while standing according to an embodiment of the invention. It can be seen that fig. 8 (a) is a total change of the three-axis acceleration (g_value), from 1G to 6G, and the magnitude of the change can be proportionally deduced from the force or severity of the fall; meanwhile, as can be seen from fig. 8 (B), the forward inclination angle (pitch) of the upper torso is changed from 0 degrees to minus 56 degrees, which means that the upper torso is inclined backward, and as can be seen from fig. 8 (C), the height of the safety symbol (i.e., the air pressure height pressure) is increased from 97280Pa to 97310Pa, which means that the height is decreased, so the state machine of fig. 8 (D) determines that the upper torso is changed from standing (state 2) to falling (state 5) to lying (state 3).

Referring to fig. 9, an embodiment of fall left detection at standing according to an embodiment of the invention. It can be seen that fig. 9 (a) shows the total change of the three-axis acceleration (g_value), from 1G to 3.5G, and the magnitude of the change can be proportionally deduced from the force or severity of the fall; meanwhile, as can be seen from fig. 9 (B), the swing angle (roll) of the upper torso is changed from 0 degrees to minus 80 degrees, which means that the upper torso is tilted to the left, and as can be seen from fig. 9 (C), the height of the safety symbol (i.e., the air pressure height pressure) is increased from 97215Pa to 97235Pa, which means that the height is decreased, so the state machine of fig. 9 (D) determines that the upper torso is changed from standing (state 2) to falling (state 5) to lying down on the left (state 8).

By the examples of fig. 8 and 9, and the related falling experiments (not all experimental values are fully shown) of the present invention, including falling in any direction, falling in the right side, falling in the left side, and by the nine-axis IMU and altimeter fixed on the safe symbol in front of the chest, and the algorithm of the finite state machine, it can be precisely determined whether the falling is generated by standing or sitting, walking or running, the falling direction, the falling force, what posture is maintained after falling, or moving after falling, even sitting or standing up, etc., compared with the conventional technology, the falling can only be known, and the invention only has obvious progress in falling detection.

In some embodiments, the peace sign may detect the quality of sleep, primarily the number and frequency of sleep posture changes, sleep posture back, prone, right side sleep, left side sleep. During sleep, the frequency of breathing and whether breathing is stopped; because the safety sign is fixed on the chest, the chest breathing fluctuation is the largest, and the IMU with high sensitivity and low noise and low drift is added, the breathing condition can be accurately judged. In addition, in sleeping, the user can easily judge whether he or she is sitting down or sitting up or sitting down or standing up, and if he or she goes to the toilet, he or she can know whether he or she moves quickly, which is the possibility of falling down easily. In addition, during sleep, if the heart beat, blood pressure and blood oxygen are abnormal, poor sleep quality may be displayed, and related diseases may be represented.

The actual application of the behavior recognition system based on the IOT positioning and wearing device provided by the invention is fully described below from a home care system.

The complete care system is built through lightweight and low cost equipment. The intelligent bracelet and the intelligent safe symbol are the only equipment to be worn, and other equipment is fixedly placed, so that the problem of excessive wearing of the sensor is solved, and meanwhile, the high accuracy of data is guaranteed.

The intelligent bracelet can detect daily physiological data such as heart rate, blood pressure, body temperature, step number, mileage, calories and the like of a subject; the intelligent security symbol can collect the data of the posture, the motion change, the indoor positioning and the like of the subject. Since the development of the smart band on the market is quite perfect, the measured physiological data is quite comprehensive, and the universality of the future experimental process is considered, the smart security symbol can be designed to be capable of receiving the data of the broadcast packet of the smart band on the market. Meanwhile, as the functions of the intelligent bracelet are developed perfectly, the function of not adding the intelligent bracelet is selected when the intelligent security symbol is developed.

The smart security symbol receives the smart bracelet data and merges the smart bracelet data and transmits the smart bracelet data to an edge device, and the edge device transmits the data to a cloud database (firestore) of Google Cloud Platfom (GCP) through an MQTT to store and deploy web pages, and simultaneously stores the web pages in a local database (MongoDB). Classifying, organizing and analyzing the data stored in the Firestore through the cloud function provided by the GCP, and displaying the data analysis result on a webpage in real time so that the subject can check the current physiological data; the data of the local database is used as a data set for training a machine learning classifier in the future, so that the value of the data is furthest exerted. And the analyzed result is actively pushed and broadcasted through the intelligent sound box by using the data stream integration function of the cloud platform.

In order to realize the function of actively pushing messages by a loudspeaker, the system is realized by using the triggering Functions of the Cloud Functions and the Cloud Pub/Sub, when data change, objects are sent to corresponding topics through HTTP triggering the Cloud Functions, the programs in the raspberry group subscribe to the same topic, text contents in the objects are received and processed through Google Text to SpeechAPI, the text contents are converted into voice messages, and finally the voice messages are actively pushed through a Google Speaker. When the intelligent security symbol detects that the subject encounters an emergency, the sound box can make emergency broadcasting through the serial connection of the data streams. Because the intelligent loudspeaker is placed in the family of the testee, and privacy doubts of the testee are considered, the microphone radio system can be started only when the intelligent loudspeaker box hears the wake-up word. Under the condition that no wake-up word is mentioned, the personal privacy of the subject is ensured.

The number of subjects was 10, 5 elderly with ages 70 to 80 and 5 young with ages 24 to 26, respectively, and the system was set in 5 families, and the young had no disease or chronic disease, and the height, weight and disease of the elderly are shown in table 6. The height values shown in Table 6 were measured without humpback. The purpose of this experiment was to gather comparative data to facilitate future training of frailty classifiers. All data are transmitted to the cloud database and the local database through the edge device. The relevant content of the experiment is collected by the daily data agreed by the inventor and accepted by the experiment.

TABLE 6 basic information Table for subject (elderly)

During the experiment, the subjects were asked to wear the smart band and the smart safe character for 7 days. Except for the toilet, the user can wear the toilet for other time. The intelligent bracelet wearing mode is characterized in that the intelligent safe symbol sticking position is located at the chest, the purpose of using sticking is to control data collection correctness, the action of the amulet is consistent with the body action, and the data interpretation of the nine-axis sensor is not affected. If the hanging mode is used for wearing, intelligent safety symbol shaking can be caused, noise is further generated, and interpretation of the sensor is affected.

The nine-axis sensor in the intelligent security symbol comprises a three-axis accelerometer, a three-axis gyroscope and a three-axis magnetometer, and can measure acceleration, rotation angle and geomagnetic direction respectively, and calculate the posture state and moving distance of the subject. And calculating the G value (1) by the square sum of the triaxial acceleration and the open root number according to the x axis, the y axis and the z axis. Whether the subject is active or not can be determined from the change in the G value, which is equal to the gravitational acceleration value (about 1G) if the subject is in a stationary state.

Through a directional algorithm (gradient descent algorithm) proposed by Madgwick, the value obtained by calculating the acceleration count value and the magnetic force count value is subjected to algorithm fusion with the value obtained by integrating the gyroscope, so that three-axis rotation angles with higher precision can be obtained, and the three-axis rotation angles are Roll, yaw, pitch respectively.

A Timed Up and Go Test (TUG) frailty assessment standard was performed to view the motor changes of the subject, and to determine the presence or absence of frailty symptoms in the subject. Time Up and Go Test the standard TUG protocol was used and 3m was calculated from the center point of the foot forward, and the cross was taped at three meters and turned around the cross while turning. TUG experiments all used chairs without back rest.

In the experimental process, the subjects wear the intelligent safety symbol and the intelligent bracelet to collect physiological state data and experimental posture data. And finally, comparing the test results and the athletic performance data of the young and the elderly, analyzing the difference and training a classifier model.

The collected data during the experiment are uploaded and stored to the cloud database through the edge device, and the current posture change is displayed in real time through the webpage for the subject to check in real time. In the future, the system development perfects and caregivers can check the activity condition of the old in real time through the webpage end. Aiming at the old people with weak symptoms, caregivers can check whether falling situations occur in real time, if the old people are unfortunately falling, the falling gesture of the intelligent peace sign can be transmitted to the cloud platform to trigger the sound box to broadcast falling warning information, and the defect caused by negligence is avoided through picture warning and broadcasting warning.

Physiological data collected daily and changes in posture are plotted for analysis. The physiological data change and posture distribution of the subject at a certain time point of the day within one hour can be clearly analyzed through the chart, and the physiological data change and posture distribution are shown in fig. 10 and 11.

From the blood pressure change within one hour in fig. 10C, it can be seen that the blood pressure change in the first 30 minutes has a tendency to occur in the pre-hypertension, and the diastolic pressure is 120mmHg to 129mmHg and the systolic pressure is lower than 80mmHg, which are the pre-hypertension standards according to the hypertension standards issued by the american heart association. However, 30 minutes after the time, the blood pressure state returns to normal, and the heart rate is higher for the first 30 minutes compared with the heart rate change chart of fig. 10A, and it is presumed that the physiological data shows the change due to the factor of tension emotion. At the same time, the posture distribution of fig. 11 is observed to show a long-time sitting state, and is verified with the experimental activity record, and the subject is watching the movie at the moment, so that the situation is estimated to be caused by the tension of the scenario.

TUG experiments were performed under the same conditions for young and old, and a comparison of the experimental results is shown in FIG. 12, in which the three-axis acceleration and the G-value are both magnified 512 times for easy observation and drawing on a graph.

By comparing the differences of the three-axis acceleration changes in fig. 12A and 12C, it can be observed that the amplitude of the G-value of the young person is larger than that of the old person, and it is inferred that the walking steps of the young person are larger and the walking speed is faster by comparing the amplitude and the observation experiment process. On the other hand, comparing the differences in the Z-axis acceleration, as shown in fig. 12A and 12C, the maximum amplitude of the Z-axis acceleration reaches-400, which indicates that the elderly get up and lean forward than the young, and by observing the experimental process, it can be deduced that the leg muscles of the elderly are slightly weak, and the elderly need to be guided to stand by body force, and the TUG test time of the elderly is longer than 12 seconds, and it is presumed that there is a risk of falling.

Through the combined analysis of the physiological data and the behavior gesture, the care system can more effectively monitor the living state of the aged at home. The intelligent bracelet and the intelligent security symbol are combined to form an Internet of things for data collection and preprocessing, the edge device uploads the data to a cloud platform database and a local end database for storage, and the cloud platform further analyzes the data and broadcasts the result through an intelligent sound box; the local database stores data for a long time and is used as a training data set of the classifier model in the future. The system can effectively observe the falling risk of the old, the web page is connected in series through the Firebase database, the data stored in the cloud database can change in real time on the web page, and the falling risk can be broadcasted in real time through the intelligent sound box, so that the regretta is avoided. The experimental result proves that the system combining the Internet of things with the cloud platform is an effective tool applied to daily care of the old.

The living habit of people in home has close relation with the placement of furniture. In a home environment, the placement position of large furniture is relatively fixed, and the position, posture and orientation of a user when using the furniture in the home are fixed. Thus, the wearer's behavioral events may be inferred from the current pose, position, and orientation.

As shown in fig. 13, we set 8 daily home environmental behavioral events, record magnetic force of three axes of each point in fig. 13 using nine-axis inertial sensors of the device, measure RSSI signal strength, place wearable device and Beacon bluetooth module in each area, and store the measurement result in Beacon bluetooth module of each area as fingerprint of behavioral event. When a wearer uses the device, the fingerprint package of the Beacon Bluetooth module closest to the device is scanned, and according to the position and the direction of the area, the area is positioned and the behavior event is acquired through the comparison of the fingerprint and the gesture. For example, sitting on a sofa in a living room on a television, standing in a bathroom on a toilet on urine, or lying on a bed on a ceiling may infer that the behavioral event is sleeping.

Indeed, the use of magnetometers alone to detect differences between test points in the same direction is not sufficient to determine nor readily distinguish the area in which the user is located. The RSSI signal strength between the scanning device and the Beacon module is required, the strongest signal is the nearest region, and then the magnetometer is used to distinguish faces in the region, which is an additional parameter of relative distance, so as to better distinguish the localization points in the same direction, and the gestures are used to distinguish their behavior events. For example, both the sofa (test point 1) and the television (test point 2) face in the same direction, the two positions can be distinguished by RSSI signal strength, and then whether the wearer is watching television or exercising can be distinguished by sitting and standing.

As shown in fig. 13, after the home setting process, the fingerprint is shown in table 7 below.

Finger print of 78 daily household environmental behavior events

/>

Wherein, the synthesized magnetic field strength MAGTotal is shown in formula (2):

although magnetometers are not disturbed by the magnetic field of the electrical apparatus, the magnetic field of the earth varies over time, and varies at the same location at different points in time. In addition, there may be differences due to the different body orientations of the testers to the anchor points each time. The displacement and movement angle information is then added to aid in the determination, such as the pose of each position, the direction of rotation from the current position to the next position, the number of steps to be moved, the distance moved, etc. The person wearing the smart peace sign walks around his home with a behavior trace as shown in fig. 14.

In the present invention, unless explicitly specified and limited otherwise, the terms "mounted," "connected," "secured," and the like are to be construed broadly, and may be, for example, fixedly connected, detachably connected, or integrally formed; can be mechanically or electrically connected; can be directly connected or indirectly connected through an intermediate medium, and can be communicated with the inside of two elements or the interaction relationship of the two elements. The specific meaning of the above terms in the present invention can be understood by those of ordinary skill in the art according to the specific circumstances.

In the description of the present specification, a description referring to terms "one embodiment," "some embodiments," "examples," "specific examples," or "some examples," etc., means that a particular feature, structure, material, or characteristic described in connection with the embodiment or example is included in at least one embodiment or example of the present invention. In this specification, schematic representations of the above terms should not be understood as necessarily being directed to the same embodiment or example. Furthermore, the particular features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples. Further, one skilled in the art can engage and combine the different embodiments or examples described in this specification.

While embodiments of the present invention have been shown and described above, it will be understood that the above embodiments are illustrative and not to be construed as limiting the invention, and that variations, modifications, alternatives and variations may be made to the above embodiments by one of ordinary skill in the art within the scope of the invention.

Claims

1. The utility model provides a behavior recognition system based on IOT location and wearing device, its characterized in that includes chest wearing device and with wireless connection's of chest wearing device IOT system, the inside microprocessor that has been constructed of chest wearing device and with nine inertial motion sensor, altimeter and the wireless communication module that the microprocessor is connected, the user will the fixed attachment of chest wearing device is in the chest to judge user's action, the node or the label of IOT system are installed in each region of living space, and its broadcast is given the packet information of chest wearing device includes: the longitude and latitude height of the installation position, the domain name of the installation position and the name of each key furniture in the area, the corresponding geomagnetic fingerprint and received signal strength, the corresponding longitude and latitude and the azimuth of the user when using the key furniture are preset when the node or the tag is installed; therefore, the chest wearing device can fuse the actions of the user with the information of the broadcast packet, the microprocessor is used for directly calculating in the chest wearing device to obtain accurate behavior identification of the user, and the obtained related information is broadcasted to the outside by the chest wearing device.

2. The IOT positioning and wearable device-based behavior recognition system according to claim 1, wherein the chest wearable device prevents falls by detecting various directions of falls and impact forces of falls, or by detecting standing instability or movement instability.

3. The IOT positioning and wearable device-based behavior recognition system according to claim 1, further comprising a smart wristband worn on a wrist of a user's hand, wherein the smart wristband detects a physiological signal and a biochemical signal between which behavior occurs and broadcasts the physiological signal and the biochemical signal outwards, wherein the physiological signal is selected from at least one of a heartbeat, an electrocardiogram, a body surface impedance, a body temperature, blood oxygen, or blood pressure, and the biochemical signal is selected from at least one of a blood glucose concentration, a lactic acid concentration, an alcohol concentration, a corticosteroid concentration, or a drug concentration.

4. The IOT positioning and wearable device-based behavior recognition system according to claim 1, 2 or 3, wherein the chest wearable device further comprises at least one of a camera or a recording device, the camera or the recording device can be triggered to take a photograph or record by action judgment or position judgment of the wearable device, or by a button of a user on the chest wearable device, or by external triggering to take a photograph or record by bluetooth, and the camera or the recording device is mounted inside the chest wearable device, or a wireless camera or a wireless recording device is additionally provided to be worn on the user.

5. The IOT positioning and wearable apparatus based behavior recognition system according to claim 3, wherein the chest wearable apparatus further comprises at least one of a camera or a sound recording device triggered to be activated by the smart band, a dominant hand wearing the smart band being brought into proximity with the chest wearable apparatus as a result of holding an item, the smart band being brought into proximity with the chest wearable apparatus as a result of the proximity, a received signal strength therebetween being relatively enhanced, thereby triggering photographing or sound recording.

6. The IOT positioning and wearable device-based behavior recognition system according to claim 3, further comprising a router and a server, wherein the router receives broadcast information packets of the chest wearable device and the smart band and uploads the broadcast information packets to the server, and the server can know the position and time period of occurrence of related behaviors of a user and habits of the user, and the server analyzes and records the related behaviors, further analyzes the correlation between the physiological signals and the biochemical signal changes during the occurrence of the related behaviors, or analyzes the stability and variability of life of the user, or analyzes the variability of the related behaviors, or analyzes the variability of specific activities.

7. The IOT positioning and wearable device-based behavior recognition system according to claim 1, wherein the pre-set of packet information broadcast to the chest wearable device is as follows:

s1, an APP is utilized to assist a user to establish an indoor plane modeling diagram of a living space, wherein the plane modeling diagram comprises positions of furniture in each area, and a socket position where the node is to be installed or a position where a wall of the tag powered by a battery is to be attached;

s2, installing the nodes to socket positions of each area according to an indoor plane modeling diagram; or attaching the label to the indoor wall of each area;

s3, the user wears the chest wearing device, geomagnetic fingerprints, received signal strength and longitude and latitude of each piece of furniture are set for each piece of furniture in each area, and the direction of the user when using the corresponding piece of furniture is set.

8. The IOT positioning and wearing device-based behavior recognition system according to claim 7, wherein a pedometer is further arranged inside the chest wearing device, the nine-axis inertial motion sensor and the altimeter calculate dead reckoning through the furniture and give a reliable position, when the user leaves the position, the position is a starting point, the pedometer and the microprocessor are used for judging the facing of the user, the read values are put into the dead reckoning to calculate the next position, and the geomagnetic fingerprint and the received signal intensity, the longitude and latitude and the azimuth of the user are combined, so that the chest wearing device obtains the next furniture closest to the user through fusion calculation, and accurate behavior recognition specific data of the user are obtained through the action of using the furniture.

9. The IOT positioning and wearable device-based behavior recognition system according to claim 1 or 7, wherein the detailed steps of presetting the package information broadcast to the chest wearable device are as follows:

s1, a user wears the chest wearing device and carries a handheld mobile device provided with a corresponding APP to go to a certain room;

s2, standing in front of a piece of furniture or using the piece of furniture, staying still, starting pressing setting in an APP interface of the handheld mobile device, informing the chest wearing device by the APP, broadcasting a setting mode to the APP by the chest wearing device, wherein the broadcasting content of the chest wearing device comprises the received signal intensity of a positioning broadcasting packet sent by the node or the tag, the azimuth of the chest wearing device and the geomagnetic intensity of the position of the chest wearing device, accumulating at least 20 pens after receiving the broadcasting content of the chest wearing device, averaging the received signal intensity, the azimuth and the geomagnetic intensity, adding the longitude and latitude of the furniture position, and setting the fingerprint information of the furniture position;

10. The IOT positioning and wearing device-based behavior recognition system according to claim 4, wherein the camera is a table camera, the camera is erected within a range of 80-150cm above a table, the table camera is a 3D camera, the shooting range is only a table area, as long as the user stays near the table, no matter sitting at the table, the chest wearing device transmits the table positioning information to a cloud end, the cloud end push requests the table camera to take a picture, the cloud end is uploaded, and the behavior of the restaurant matches the content of the picture to judge the activity type and the diet content of the user.