CN113749644B - Intelligent garment capable of monitoring lumbar vertebra movement of human body and correcting autonomous posture - Google Patents
Intelligent garment capable of monitoring lumbar vertebra movement of human body and correcting autonomous posture Download PDFInfo
- Publication number
- CN113749644B CN113749644B CN202110886147.5A CN202110886147A CN113749644B CN 113749644 B CN113749644 B CN 113749644B CN 202110886147 A CN202110886147 A CN 202110886147A CN 113749644 B CN113749644 B CN 113749644B
- Authority
- CN
- China
- Prior art keywords
- signal
- posture
- acceleration
- frequency domain
- xyz
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Classifications
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/103—Detecting, measuring or recording devices for testing the shape, pattern, colour, size or movement of the body or parts thereof, for diagnostic purposes
- A61B5/11—Measuring movement of the entire body or parts thereof, e.g. head or hand tremor, mobility of a limb
- A61B5/1116—Determining posture transitions
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/103—Detecting, measuring or recording devices for testing the shape, pattern, colour, size or movement of the body or parts thereof, for diagnostic purposes
- A61B5/107—Measuring physical dimensions, e.g. size of the entire body or parts thereof
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/103—Detecting, measuring or recording devices for testing the shape, pattern, colour, size or movement of the body or parts thereof, for diagnostic purposes
- A61B5/11—Measuring movement of the entire body or parts thereof, e.g. head or hand tremor, mobility of a limb
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/103—Detecting, measuring or recording devices for testing the shape, pattern, colour, size or movement of the body or parts thereof, for diagnostic purposes
- A61B5/11—Measuring movement of the entire body or parts thereof, e.g. head or hand tremor, mobility of a limb
- A61B5/1123—Discriminating type of movement, e.g. walking or running
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/68—Arrangements of detecting, measuring or recording means, e.g. sensors, in relation to patient
- A61B5/6801—Arrangements of detecting, measuring or recording means, e.g. sensors, in relation to patient specially adapted to be attached to or worn on the body surface
- A61B5/6802—Sensor mounted on worn items
- A61B5/6804—Garments; Clothes
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/74—Details of notification to user or communication with user or patient ; user input means
- A61B5/746—Alarms related to a physiological condition, e.g. details of setting alarm thresholds or avoiding false alarms
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/243—Classification techniques relating to the number of classes
- G06F18/24323—Tree-organised classifiers
Landscapes
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Public Health (AREA)
- Animal Behavior & Ethology (AREA)
- Surgery (AREA)
- Molecular Biology (AREA)
- Medical Informatics (AREA)
- General Health & Medical Sciences (AREA)
- Heart & Thoracic Surgery (AREA)
- Biomedical Technology (AREA)
- Pathology (AREA)
- Biophysics (AREA)
- Veterinary Medicine (AREA)
- Physiology (AREA)
- Oral & Maxillofacial Surgery (AREA)
- Dentistry (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- Measurement Of The Respiration, Hearing Ability, Form, And Blood Characteristics Of Living Organisms (AREA)
Abstract
The invention relates to an intelligent garment capable of monitoring lumbar vertebra movements of a human body and correcting autonomous postures, which particularly comprises a garment body, an identification monitoring module and a vibration reminding module. The system firstly obtains the real-time motion state and gesture of the user through the human motion state and gesture recognition module, and obtains the real-time reasonable threshold value of the waist bending Cobb angle of the user based on the gesture evaluation system constructed in advance. Then, the system acquires the Cobb angle of the real-time waist bending of the user through the waist movement angle measuring module, and simultaneously judges whether the angle exceeds a reasonable threshold for too long or not and decides whether to remind the user of taking the waist posture through the vibration device or not. According to the invention, the blind powerful correction method of the original product is replaced by the movement state and posture identification, vibration reminding and automatic correction in the aspect of lumbar vertebra correction, so that the discomfort of lumbar vertebra diseased people in the correction process can be effectively reduced, and the wide lumbar vertebra people can be helped to develop good lumbar using habits.
Description
Technical Field
The invention belongs to the technical field of human body posture monitoring, and particularly relates to intelligent clothing capable of monitoring and autonomously correcting human lumbar vertebra movements.
Background
With the development and popularization of the Internet, the activities such as entertainment, study and office of modern young people are obviously changed, and long-term low-head mobile phone playing and long-term living habit of sitting for a computer cause a series of lumbar vertebra health problems. There is data relating to the trend that lumbar diseases are increasingly becoming more popular and younger.
Most young individuals use an auxiliary corrective belt known as a lumbar fixator. The waist movement of the instrument is restrained by a powerful fixing mode to prevent the occurrence of unreasonable waist use. However, this correction method is extremely likely to cause a user's sense of rejection due to problems such as excessive binding to the waist, poor wearing comfort, and adverse physical activities. Moreover, long-term use of such devices by patients may stress blood vessels and nerves, which are not conducive to blood reflux and may cause some degree of nerve damage.
In the field of intelligent medical treatment, human posture correction devices based on human posture recognition are also layered endlessly. The device is mainly characterized by being capable of recognizing concentrated different postures of a human body and reminding. However, under the same motion state, the human body has different effects on the lumbar vertebra of the human body due to different instantaneous postures; the same instantaneous posture has different influences on the lumbar vertebrae of the human body under different motion states. Therefore, the rationality of the waist posture of the human body needs to be evaluated in combination with the motion state of the human body and the instantaneous posture in the motion state. Secondly, some correction devices currently exist only including a reminding function, and no specific scheme is provided in terms of how to correct, adjust, etc. In addition, the existing correcting device as a device cannot be worn in daily work and study of a user, is large in size and lacks in aesthetic property.
Therefore, in relieving and treating lumbar diseases which are increasingly popular and younger, researchers need to consider not only the effectiveness problem of the lumbar auxiliary correction device, but also the correction rationality, wearing comfort and negative effects caused in the correction process under different motion states and different postures, and the problems of normal wearing and use in daily work and study of users.
The invention comprises an ergonomic flexible correction function and a human motion state and gesture recognition function, and can make up for the defects of the traditional lumbar auxiliary correction equipment in the aspects to a certain extent through a flexible correction and vibration reminding mode, and enables a user to automatically develop good waist habit.
Disclosure of Invention
The invention aims to provide intelligent clothing capable of monitoring lumbar vertebra movements of a human body and correcting the autonomous postures.
The intelligent garment comprises a garment body, an identification monitoring module and a vibration reminding module, wherein the identification monitoring module and the vibration reminding module are arranged on the garment body, the identification monitoring module is used for collecting the motion state and body posture information of a human body, a human body life posture evaluation model is built, a reasonable value range of a Cobb angle under the current motion state and posture is obtained through the human body life posture evaluation model, the intelligent garment is further used for calculating the real-time waist bending Cobb angle of a user by using a posture evaluation system, judging whether the Cobb angle is in a reasonable value range, and if not, starting the vibration reminding module, the vibration reminding module is used for reminding the user to pay attention to the waist posture by vibration.
Further, the recognition monitoring module comprises a depth sensor and a triaxial angular acceleration sensor, the depth sensor comprises a right shoulder joint depth sensor, a left elbow joint depth sensor and a right elbow joint depth sensor which are embedded in the positions of the arm and arm joints of the garment, the depth sensor all uses an ultrasonic ranging principle to obtain the relative distance between the garment joints in a crossing manner, the triaxial angular acceleration sensor comprises a first triaxial angular acceleration sensor arranged at the rear collar of the garment body, a second triaxial angular acceleration sensor embedded in the first lumbar vertebra of the garment body and a third triaxial angular acceleration sensor embedded in the fifth lumbar vertebra of the garment body, the first triaxial angular acceleration sensor is used for recognizing the motion state of the human body, and the second triaxial angular acceleration sensor and the third triaxial angular acceleration sensor are respectively used for obtaining the vertical ground angle acceleration of the first lumbar vertebra and the fifth lumbar vertebra so as to monitor the physiological curvature of the human body in real time.
Further, the real-time waist bending Cobb angle of the user is calculated by respectively measuring the vertical earth angle alpha at the first lumbar vertebra and the vertical earth angle beta at the fifth lumbar vertebra in a mode of integrating angular acceleration with time; and subtracting the vertical earth angle beta at the fifth lumbar vertebra from the vertical earth angle alpha at the first lumbar vertebra to obtain a difference value between beta and alpha, wherein the difference value is the lumbar vertebra Cobb angle measured by the sensor.
Further, the waistband can be dismantled and set up on the clothing body, and the waistband includes waistband body and composite layer, the composite layer sets up at the waistband middle part, the composite layer includes from outside to interior composite material overburden, first flexible buffer layer, medical nylon supporting layer and the flexible buffer layer of setting in order.
Further, the method for establishing the human body life posture evaluation system comprises the following steps:
step 1, a first triaxial acceleration sensor acquires 3-axis original signals tac-XYZ and tGyro-XYZ of an accelerometer and a gyroscope;
step 2, a median filter and a three-order low-pass Butterworth filter with the angular frequency of 20Hz are used for filtering tac-XYZ and tGyro-XYZ, and the other low-pass Butterworth filter with the angular frequency of 0.3Hz is used for dividing an acceleration signal tac-XYZ into a body acceleration signal body Acc-XYZ and a gravity acceleration signal tGravityAcc-XYZ;
step 3, deriving the linear acceleration and the angular velocity of the body, and obtaining reflected signals, namely a body acceleration time domain reflected signal tBeodyAccJerk-XYZ returned by an accelerometer and a body angular velocity time domain returned signal tBeodyGyro Jerk-XYZ returned by a gyroscope, calculating the amplitude of the three-dimensional signals by using Euclidean norms, wherein the three-dimensional signals are a body acceleration time domain amplitude signal tBeodyAccMag, a gravity acceleration time domain amplitude signal tGravityAccMag, a body acceleration time domain amplitude reflected signal tBeodyAccJerkMag, a body angular velocity time domain amplitude signal tBeodyGyroMag and a body angular velocity time domain amplitude reflected signal tBeodyGyro JerkMag respectively;
Step 4, performing fast Fourier transform on the generated body acceleration frequency domain signal, body acceleration frequency domain reflection signal, body angular velocity frequency domain signal, body acceleration frequency domain amplitude reflection signal, body angular velocity frequency domain amplitude signal and body angular velocity frequency domain amplitude reflection signal, obtaining a body acceleration time domain signal tbodyoacc-XYZ, a gravity acceleration time domain signal tgovityacc-XYZ, a body acceleration time domain reflection signal tbodyojerk-XYZ, a body angular velocity time domain signal tBodyGyro-XYZ, a body angular velocity time domain reflection signal tbodygoro-XYZ, a body acceleration time domain amplitude signal tbodyoacmag, a gravity acceleration time domain amplitude signal tgovityacmag, a body acceleration time domain amplitude reflection signal tbodyojerkmag, a body angular velocity time domain amplitude signal tBodyGyroMag, a body angular velocity time domain amplitude reflection signal tbodygojerkmag, a body acceleration frequency domain signal fbodyork-XYZ, a body acceleration frequency domain reflection signal fbodygork-XYZ, a body angular velocity frequency domain signal fbodygorq-XYZ, a body acceleration frequency domain amplitude signal fbodygorkmag, a body angular velocity frequency domain amplitude signal tbodygorkmag and a body angular velocity frequency domain amplitude signal tbodygorkmag, wherein the body angular velocity frequency domain amplitude signal tbodygorkMag and jkz are used for representing a body direction of a 'and a' direction of a '62' of a body;
Step 5, obtaining a time domain feature vector and a frequency domain feature vector through the signals in the step 1-3, and recording the motion state information of the human body in different time domains and frequency domains;
and 6, classifying 6 basic motion states of standing, sitting, prone, walking, going upstairs and downstairs of the human body by a hierarchical classification method, taking data containing time domain feature vectors and frequency domain feature vectors as a training set in each hierarchy, and inputting the training set into a random forest classification model for training to obtain a human body life posture evaluation model.
Further, the hierarchical classification method specifically includes the following steps:
step 6.1, classifying static gestures and dynamic gestures in the 1 st level, firstly, uniformly defining a target prediction set corresponding to a sample with a motion state of standing, sitting and prone as a static gesture and datating the static gesture as the same value; the method comprises the steps that a target prediction set corresponding to a sample with a motion state of walking, going upstairs and going downstairs is uniformly specified as a dynamic gesture and is dataized to be another same value, then the dataized target prediction set is subjected to data prediction by using a random forest algorithm to obtain a prediction classification result, and the level 2 is entered after the prediction classification is completed;
Step 6.2, classifying walking and going upstairs and downstairs in the dynamic gestures in the 2 nd level, screening out samples with the target prediction set being the dynamic gesture in the original data set, and uniformly prescribing the target prediction set corresponding to the samples with the dynamic gesture being the walking gesture as the walking gesture and digitizing the walking gesture; uniformly defining target prediction sets corresponding to upstairs and downstairs samples in the dynamic gestures as upstairs and downstairs gestures, and datating the target prediction sets to be the same value, finally, carrying out data training by using a random forest algorithm, wherein the training process is the same as that in the step 6.1, and entering a 3 rd level after the prediction classification is finished;
step 6.3, classifying upstairs and downstairs in upstairs and downstairs postures at the 3 rd level, screening out samples with target prediction sets of upstairs or downstairs in the original data set, inputting the samples of upstairs or downstairs into a random forest algorithm for prediction classification, and entering the 4 th level after the prediction classification is finished;
step 6.4, classifying standing postures and sitting postures in the static postures by the 4 th hierarchy, firstly screening samples with static posture as target prediction sets in the original data set, and then uniformly prescribing the target prediction sets corresponding to the samples with the static postures as standing postures and datating the target prediction sets as the same value; uniformly prescribing a target prediction set corresponding to a standing posture and a sitting and lying sample in the static posture as a sitting and lying posture and datating the sitting and lying posture as another same value; then, carrying out data prediction on the target prediction set after the data processing by using a random forest algorithm to obtain a prediction classification result, and entering a 5 th level after the prediction classification is finished;
And 6.5, classifying sitting postures and prone postures in sitting and lying at the 5 th level, screening out samples of which the target prediction set is the sitting postures or the prone postures in the original data set, inputting the samples of the sitting postures and the prone postures into a random forest algorithm for prediction classification, obtaining a final prediction classification result, finishing the data training process by the prediction classification, and obtaining a human life posture evaluation model.
Further, the mode of obtaining the reasonable value range of the Cobb angle under the current motion state and the gesture by the human life gesture evaluation model is that after the current motion state and the gesture are obtained, the reasonable value range of the Cobb angle under the current motion state and the gesture is obtained according to the current medical standard.
Further, the method for predicting data by using the random forest algorithm comprises the following steps:
step 6.11, carrying out replaceable random sampling on a training set containing data of time domain feature vectors and frequency domain feature vectors, randomly taking out m feature vectors each time, and sampling for n times;
step 6.12, generating 1 decision tree for the feature vector matrix obtained by sampling each time, and generating n decision trees;
step 6.13, based on calculation and judgment modes of the information gain and the information gain rate of the decision tree, an algorithm forms n decision tree prediction models and predicts a target prediction set of the sample;
And 6.14, determining a final prediction classification result in a manner of voting n decision trees.
The beneficial effects of the invention are as follows: 1. in the process of realizing the correction function, the invention adopts a mode of coexistence of moderate correction and autonomous correction. The vibration reminding of the system is combined with the flexible stationary phase of the clothing waistband to help the user to correctly use the waist and develop good waist-using habit. 2. In the invention, in the recognition of the waist gesture of the human body, the mode of combining the human body motion state recognition and the human body gesture recognition is adopted, the precision of the waist gesture recognition is improved, and the correction rationality of the garment in the use process is ensured. 3. In the measurement of the physiological curvature of the lumbar vertebra of the human body, the adopted system adopts the Cobb angle as a main index for monitoring the physiological curvature of the lumbar vertebra of the human body. The finger has strong correlation with the physiological curvature of the lumbar vertebra of the human body in standard medicine and is easier to measure. Simplifying the process of rationality evaluation of the physiological curvature of the lumbar vertebra of the human body. 4. The Bluetooth data transceiver module can upload the measured Cobb angle data to the mobile terminal in real time. By analyzing the data, the daily waist habit information of the user can be obtained, and the severity degree and recurrence probability of the disease can be estimated. 5. The garment is used as a carrier, and the human body motion state recognition system, the human body posture recognition system, the human lumbar physiological curvature monitoring system and the signal receiving and transmitting module are all arranged in the intelligent garment, so that a user can conveniently carry the garment in daily work and study, and aesthetic requirements of the user in daily activities are met. 6. The human body life posture evaluation model is independently designed, so that the human body posture can be effectively classified, and the accuracy is high.
Drawings
FIG. 1 is a schematic front view of a human lumbar motion monitoring and autonomous posture correction smart garment of the present invention;
FIG. 2 is a schematic back view of the intelligent garment for human lumbar motion monitoring and autonomous posture correction of the present invention;
FIG. 3 is a schematic diagram of the front waist shape of the intelligent garment for human lumbar motion monitoring and autonomous posture correction according to the present invention;
FIG. 4 is a schematic view of the back waist shape of the intelligent garment for human lumbar motion monitoring and autonomous posture correction according to the present invention;
FIG. 5 is a schematic diagram of the internal structure of the waist of the intelligent garment for human lumbar motion monitoring and autonomous posture correction according to the present invention;
fig. 6 is a schematic diagram of lumbar vertebra bending angle measurement of the intelligent garment for human lumbar vertebra motion monitoring and autonomous posture correction according to the present invention;
FIG. 7 is a flow chart of the functional implementation of the intelligent garment for human lumbar motion monitoring and autonomous posture correction of the present invention;
FIG. 8 is a flow chart of a human motion state recognition method of the present invention;
FIG. 9 is a flowchart of a hierarchical classification method for human motion state identification according to the present invention;
FIG. 10 is a diagram showing a data distribution of each motion state in a human motion state recognition training set according to the present invention;
FIG. 11 is a graph showing the data distribution of each movement state in the human movement state recognition test set according to the present invention;
FIG. 12 is a graph showing a data distribution of feature vectors angle (X, gradyMean) of various motion states in a time domain in a test set for human motion state recognition according to the present invention;
FIG. 13 is a diagram showing the data distribution of feature vectors tGravityAcc-min () -X of each motion state in a time domain in a test set of human motion state recognition according to the present invention;
FIG. 14 is a diagram showing the data distribution of the feature vector tBedyAcc-energy () -X of each motion state in a time domain in the test set for human motion state identification according to the present invention;
FIG. 15 is a graph showing the data distribution of feature vectors TBodyAcc-mean () -X for each motion state in a time domain in a test set for human motion state identification according to the present invention;
FIG. 16 is a graph showing the clustering results of the intelligent garment for human lumbar motion monitoring and autonomous posture correction according to the present invention, wherein the high-dimensional manifold is mapped to two-dimensional space by T-SNE manifold learning;
FIG. 17 is a graph showing the accuracy test results of the intelligent garment for human lumbar motion monitoring and autonomous posture correction according to the present invention for different motion state test sets;
fig. 18 is a graph showing the accuracy test results of the intelligent garment for human lumbar motion monitoring and autonomous posture correction according to the present invention for different posture test sets of the human body.
The list of components represented by the various numbers in the drawings is as follows:
30. a garment body; 1. a right shoulder joint depth sensor; 2. a left shoulder joint depth sensor; 3. left elbow joint depth sensor; 4. a right elbow joint depth sensor; 5. the right waist is fixed with a fastening tape surface; 6. the left waist part is fixed with a fastening tape surface; 7. a first axis angular acceleration sensor; 8. a waistband body; 9. a second triaxial angular acceleration sensor; 10. a third triaxial angular acceleration sensor; 11. a power switch; 12. the left waist is fixed with the female surface of the sticking buckle belt; 13. the right waist is fixed with the female surface of the sticking buckle belt; 14. a waistband ventilation layer; 15. a waistband composite cover layer; 16. a waistband flexible buffer layer A; 17. a medical nylon supporting layer; 18. a flexible buffer layer B; 19. a micro-pressurization vibration unit; 20. a flexible signal transmission line; 21. a flexible main control circuit; 22. a Bluetooth data receiving and transmitting module; 23. a storage battery; 24. a first lumbar vertebra; 25. a fifth lumbar vertebra; 26. a vertical ground-to-ground angle alpha at the first lumbar vertebra; 27. a vertical ground-to-ground angle beta at the fifth lumbar vertebra; 28. the difference between beta and alpha; 29. lumbar Cobb angle.
Detailed Description
The invention is described in further detail below with reference to the attached drawings and specific examples:
Fig. 1 shows a schematic front view of a garment body 30 according to the present invention. The right shoulder joint depth sensor 1, the left shoulder joint depth sensor 2, the left elbow joint depth sensor 3 and the right elbow joint depth sensor 4 are respectively embedded in the arm joints and the arm joints of the garment. The ultrasonic ranging principle is adopted by the depth sensors everywhere, so that the relative distance between the clothing joints can be obtained across obstacles. The front waist of the garment is respectively provided with a right waist fixing thread gluing belt surface 5 and a left waist fixing thread gluing belt surface 6 which are used for fixing the waistband body 8.
Fig. 2 shows a schematic back view of a garment body 30 of the present invention. The first triaxial angular acceleration sensor 7 is embedded in the rear collar of the clothing body 30, and the sensor model is mpu6050. The mpu6050 sensor can acquire angular velocity data and acceleration data of a human body in three directions of X, Y, Z axes. The waist of the clothing body 30 is provided with a waistband body 8 which accords with human engineering and is used for slightly correcting the posture of the waist of the human body and monitoring the physiological curvature of the lumbar vertebra in real time.
Fig. 3 shows a schematic front view of the waistband body 8 according to the present invention. The left and right sides of the front of the waistband body 8 are respectively provided with a left waist fixing thread gluing tape female surface 12 and a right waist fixing thread gluing tape female surface 13. The two sides of the middle part of the waistband body 8 are provided with waistband air permeable layers 14 for ensuring good air permeability of the waistband.
Fig. 4 shows a schematic back view of the waistband body 8 according to the present invention. The outermost layer of the middle part of the waistband body 8 is a waistband composite material cover layer 15. The covering layer conforms to the contour of the waist structure of the human body and can be tightly attached to the skin of the waist of the human body. The second layer in the middle of the waistband body 8 is a waistband flexible buffer layer A16, and plays a role in relieving waistband constraint and human body waist pressure together with the waistband flexible buffer layer B18 in the innermost layer in the middle of the waistband body. The third layer in the middle of the waistband body 8 is a medical nylon supporting layer 17 which is used for properly correcting the posture of the waist.
Fig. 5 is a schematic view showing the internal structure of the waistband body 8 according to the present invention. The upper side and the lower side of the central axis of the inner structure of the waistband body 8 are respectively embedded with a second triaxial angular acceleration sensor 9 and a third triaxial angular acceleration sensor 10, and the model is mpu6050. The upper triaxial angular acceleration sensor and the lower triaxial angular acceleration sensor are respectively used for acquiring the vertical earth angular acceleration of the first lumbar vertebra and the fifth lumbar vertebra. The middle part of the central axis of the internal structure of the waistband body 8 is embedded with a micro pressurizing vibration unit 19. The unit can be used for giving an unreasonable waist-use reminder to the user in a vibrating manner. The belt body 8 has a flexible signal transmission line 20 connecting the micro-pressurization vibration unit 19 and the flexible main control circuit 21. The signal transmission line adopts a Dupont line and is used for signal transmission between the second triaxial angular acceleration sensor 9, the third triaxial angular acceleration sensor 10, the micro-pressurization vibration unit 19 and the flexible main control circuit 21. The flexible main control circuit 21 comprises a singlechip with a specific structure and is used for controlling various electronic elements of the waistband. The left side of the flexible main control circuit 21 is connected with the power switch 11 and the Bluetooth data transceiver module 22. The Bluetooth data transceiver module can send monitoring data to the mobile terminal and receive various instructions of the mobile terminal. A storage battery 23 is connected below the flexible main control circuit 21. The battery 23 can be charged and discharged for powering the flexible main control circuit.
Fig. 6 is a schematic diagram of lumbar vertebra bending angle measurement of the waistband body 8 according to the present invention. The first lumbar vertebra 24 and the fifth lumbar vertebra 25 are somewhat attached to the second triaxial angular acceleration sensor 9 and the third triaxial angular acceleration sensor 10, respectively. The upper and lower sensors can measure the vertical earth angle acceleration of the first lumbar vertebra 24 and the fifth lumbar vertebra 25 respectively, and measure the vertical earth angle alpha 26 at the first lumbar vertebra and the vertical earth angle beta 27 at the fifth lumbar vertebra respectively by integrating the angular acceleration with time. The difference 28 between beta and alpha can be obtained by the algorithm by subtracting the vertical earth angle beta 27 at the fifth lumbar vertebra from the vertical earth angle alpha 26 at the first lumbar vertebra, and the difference is taken as the lumbar Cobb angle 29 measured by the sensor.
The specific implementation process of the functions of the invention is as follows:
as shown in fig. 7, the present invention firstly uses the first triaxial angular acceleration sensor 7 embedded in the rear collar of the garment body 30 to obtain angular velocity and acceleration data of the human body in the X, Y, Z three axial directions, and obtains corresponding angular and acceleration derivative data through a certain data processing (the obtaining of the derivative data will be specifically described in the further description of the technical scheme of the present invention). And carrying out data training on the data by using a random forest algorithm to finally obtain a corresponding human motion state prediction model. Based on the model, the system may identify real-time motion states of the human body. The present invention then makes it possible to obtain the relative distance of the sensors around by means of the depth sensors 1, 2, 3, 4 of the garment body 8 at the joints around. The system adopts a coordinate method and a distance method and utilizes a convolutional neural network algorithm to carry out data training on the relative distance data so as to construct a corresponding human body posture prediction model. Based on the model, the system can recognize the real-time gesture of the human body, and finally obtains the specific gesture of the user in the specific motion state. After the motion state and posture information of the human body are acquired, the system acquires a reasonable value range of the Cobb angle under the current motion state and posture through a human body life posture evaluation system established by a cloud database. ( Under different motion states and postures, the lumbar anterior lobe (Cobb angle) of people of different sexes and ages has a conventional and reasonable standard in medicine, and the human life posture evaluation system is established on the medical standard. The specific medical criteria are shown in Table 1 (which lists the reasonable value range of lumbar anterior lobe in only a partial posture). )
Table 1 average lumbar anterior lobe (Cobb angle) measured under different conditions
* Bending buttocks and knees.
1 When performing vertical Magnetic Resonance Imaging (MRI), the subject is required to lean slightly on the table and place the arm on the rail to ensure inactivity.
Meanwhile, the invention utilizes the second triaxial angular acceleration sensor 9 and the third triaxial angular acceleration sensor 10 on the upper side and the lower side of the central axis in the waistband body 8 to measure the Cobb angle of the human body at the moment and judge whether the angle is within a reasonable threshold value. When the measured Cobb angle is in a reasonable range, the micro pressurizing vibration unit 19 at the central axis of the waistband body 8 does not vibrate; when the measured Cobb angle exceeds a reasonable range and exceeds a long time, the micro-pressurization vibration unit 19 vibrates and prompts the user to adjust the waist posture autonomously. In addition, the waistband body 8 of the invention comprises a flexible correcting function, and can properly correct the waist posture through the medical nylon supporting layer 17. Meanwhile, the Bluetooth data receiving and transmitting module in the internal structure of the waistband body 8 sends the Cobb angle data of the user to the mobile terminal in real time, so that the real-time monitoring of the physiological curvature of the lumbar vertebra of the human body is realized
The technical scheme of the invention is further described below.
As shown in fig. 8, the acquisition of angular acceleration derived data and the identification of human motion state according to the present invention comprises the following steps:
first, a three-axis acceleration sensor embedded in the back collar of the garment acquires the original signals tac-XYZ and tGyro-XYZ of the accelerometer and gyroscope 3 axes. These time domain reflected signals (prefix "t" denotes time) are captured at a constant frequency of 50 hz.
The system was then filtered using a median filter and a third order low pass butterworth filter with an angular frequency of 20hz to remove noise. Meanwhile, the system uses another low-pass Butterworth filter with an angular frequency of 0.3Hz, and the acceleration signal is divided into body and gravitational acceleration signals (body Acc-XYZ and tGravityAcc-XYZ).
The system then derives the body linear and angular acceleration in time, obtaining reflected signals, namely the body acceleration time domain reflected signal returned by the accelerometer (tbodyojerk-XYZ) and the body angular velocity time domain returned signal returned by the gyroscope (tbodygryjerk-XYZ). The euclidean norms are used to calculate the amplitudes of these three-dimensional signals. The amplitude is expressed as a body acceleration time domain amplitude signal (tBodyAccMag), a gravitational acceleration time domain amplitude signal (tgavityaccmag), a body acceleration time domain amplitude reflected signal (tBodyAccJerkMag), a body angular velocity time domain amplitude signal (tBodyGyroMag) and a body angular velocity time domain amplitude reflected signal (tBodyGyroJerkMag).
Finally, the system performs a Fast Fourier Transform (FFT) on the generated body acceleration frequency domain signal, body acceleration frequency domain reflected signal, body angular velocity frequency domain signal, body acceleration frequency domain amplitude reflected signal, body angular velocity frequency domain amplitude reflected signal (i.e., fbody acc-XYZ, fBodyAccJerk-XYZ, fbody gyro-XYZ, fBodyAccJerkMag, fBodyGyroMag, fBodyGyroJerkMag). ("f" represents a frequency domain signal). These signals can be used to estimate the variance of the feature vector in each mode. These signals include body acceleration time domain signal tBeodyAcc-XYZ, gravity acceleration time domain signal tGravityAcc-XYZ, body acceleration time domain reflection signal tBeodyAccJerk-XYZ, body angular velocity time domain signal tBeodyGyro-XYZ, body angular velocity time domain reflection signal tBeodyGyr-XYZ, body acceleration time domain amplitude signal tBeodyAccMag, gravity acceleration time domain amplitude signal tGravityAccMag, body acceleration time domain amplitude reflection signal tBeodyAccJerkMag, body angular velocity time domain amplitude signal tBeodyGyrkMag, body angular velocity time domain amplitude reflection signal tBeodyGyrkMag, body acceleration frequency domain signal fBodyAcc-XYZ, body angular velocity frequency domain reflection signal fBodyAccJerK-XYZ, body angular velocity frequency domain signal fBodyGyro-XYZ, body acceleration frequency domain amplitude signal fBodyAccMag, body angular velocity frequency domain amplitude signal BodykMag, body angular velocity frequency domain amplitude signal BodyMag, body angular velocity frequency domain reflection signal BodykMag. Wherein "—xyz" is used to represent X, Y and the 3-axis signal in the Z direction.
The variable set obtained by the system through the signals comprises time domain features such as entropy, rotation angle, elevation angle, average value, absolute deviation, quartile, kurtosis, median, standard deviation and variance, and frequency domain features such as energy, entropy and DC mean value. Finally, the system obtains 561 eigenvectors containing time domain and frequency domain variables, and records the motion state information of the human body under different time domains and frequency domains.
Finally, the system adopts a hierarchical classification method to classify 6 basic motion states of the human body. In each layer, 561 feature vector data containing time domain and frequency domain variables are used as training sets and input into a random forest classification algorithm for data training, and a human motion state recognition model is obtained.
As shown in fig. 9, the hierarchical classification process in the human motion state of the present invention comprises the following steps:
the system initially divides the human motion state into 6 basic gestures, including 3 static gestures (standing gesture, sitting gesture, prone gesture) and 3 dynamic gestures (walking, going upstairs and going downstairs), and divides the recognition process into 5 layers.
The 1 st hierarchy is used to classify static and dynamic gestures. Firstly, uniformly prescribing a target prediction set corresponding to a sample with a motion state of standing posture, sitting posture and prone posture as a static posture and datating the static posture as the same value; the target prediction set corresponding to the samples with the motion states of walking, going upstairs and going downstairs is uniformly defined as a dynamic gesture and is dataized into another same value. And then training data by using a random forest algorithm. The data training process is that 561 eigenvectors are sampled randomly and m eigenvectors are taken out randomly each time and sampled n times. And then generating 1 decision tree for the eigenvector matrix obtained by each sampling, and generating n decision trees. Based on the calculation and judgment modes of the information gain and the information gain rate of the decision tree, an algorithm forms n decision tree prediction models and predicts a target prediction set of the sample. And finally, determining the value of the target prediction set of the final sample in a mode of voting n decision trees. And entering a 2 nd layer after the prediction classification is completed.
The 2 nd level is used to classify walking and going upstairs and downstairs in dynamic gestures. First, a sample with a target prediction set being a dynamic gesture in an original data set is screened out. Then, uniformly prescribing a target prediction set corresponding to a sample with a dynamic posture of walking as a walking posture and carrying out datamation (the datamation is the same value); the target prediction set corresponding to the samples for going upstairs and downstairs among the dynamic gestures is uniformly defined as the upstairs and downstairs gestures and is data as the same value. And finally, carrying out data training by using a random forest algorithm, wherein the training process is the same as that of the data training. And entering a 3 rd level after the prediction classification is completed.
The 3 rd level is used for classifying upstairs and downstairs among upstairs and downstairs. First, a sample with a target prediction set of upstairs and downstairs in an original data set is screened out. Then, because the upper building and the lower building are respectively dataized and have the same value, the random forest algorithm data training can be directly carried out, and the training process is the same as that of the upper building. After the prediction classification is completed, the 4 th hierarchy is entered.
The 4 th hierarchy is used to classify standing and sitting postures in static postures. First, a sample divided into static poses of a target prediction set in an original data set is screened. Then uniformly defining a target prediction set corresponding to a sample with a static posture as a standing posture as the standing posture and carrying out data (the data is the same value); the target prediction set corresponding to the standing posture and sitting and lying sample in the static posture is uniformly defined as the sitting and lying posture and is data as the same value. And then, carrying out data training by using a random forest algorithm, wherein the training process is the same as that of the data training. After the prediction classification is completed, the 5 th hierarchy is entered.
The 5 th level is used for classifying sitting postures and prone postures in sitting and lying postures. First, a sample in which a target prediction set in the original data set is sitting and lying is screened out. Then, because the sitting posture and the prone posture are respectively dataized and are the same value, the random forest algorithm data training can be directly carried out in the layer, and the training process is the same as that described above. And finishing the data training process by prediction classification to obtain the human motion gesture recognition model. For a newly acquired sample, the identification of the motion state type of the newly acquired sample is realized through a hierarchical classification flow.
As shown in fig. 10 and 11, 10299 pieces of data are measured in the construction of the human motion state recognition model, wherein 7352 pieces of test set and 2947 pieces of test set are used in the invention.
As shown in fig. 12, 13, 14 and 15, in the construction of the human motion state recognition model, the data distribution of feature vectors angle (X, graditymean), tgovityacc-min () -X, tBodyAcc-energy () -X, TBodyAcc-mean () -X in each motion state in a certain time domain is shown. It can be found that in this selected time domain, the feature vectors in the respective motion states have a relatively significant difference.
As shown in fig. 16, in the construction of the human motion state recognition model, the present invention maps the high-dimensional manifold to the clustering result of different motion states of the human body after two-dimensional space through T-SNE manifold learning. From the figure, other motion states are distinguished obviously except that the clustering result of sitting (motion state 3) and standing (motion state 4) is not ideal.
As shown in fig. 17, in the identification of the motion state of the human body, the accuracy of the model obtained by training the data by using the sample test set is tested. The abscissa in the figure represents the result of the algorithm prediction, and the ordinate represents the actual motion state. The diagonally coincident regions represent the correct number of samples for different motion states. The prediction accuracy of the walking state (0) is calculated to be 97.5%, the prediction accuracy of the upstairs state (1) is calculated to be 92.5%, the prediction accuracy of the downstairs state (2) is calculated to be 85.1%, the prediction accuracy of the sitting state (3) is calculated to be 89.6%, the prediction accuracy of the standing state (4) is calculated to be 97.1%, and the prediction accuracy of the lying state (5) is calculated to be 100%. The overall prediction accuracy is as high as 93.7%.
As shown in fig. 18, in the present invention, in the recognition of the human body posture, the accuracy test result of predicting 5 kinds of human body upper limb postures is performed. Wherein the gestures A, C, D, E can be effectively identified, and the identification accuracy of the gesture B is as high as 90%.
In summary, the intelligent garment for human lumbar vertebra movement monitoring and autonomous posture correction has the main characteristics that:
first, the present invention adopts a mode of coexistence of moderate correction and autonomous correction in the process of realizing correction functionality. The vibration reminding of the system is combined with the flexible stationary phase of the clothing waistband to help the user to correctly use the waist and develop good waist-using habit.
Secondly, in the recognition of the waist gesture of the human body, the invention adopts a mode of combining the human body motion state recognition and the human body gesture recognition, thereby improving the precision of the waist gesture recognition and ensuring the correction rationality of the garment in the use process.
Thirdly, in the measurement of the physiological curvature of the lumbar vertebra of the human body, the adopted system adopts the Cobb angle as a main index for monitoring the physiological curvature of the lumbar vertebra of the human body. The finger has strong correlation with the physiological curvature of the lumbar vertebra of the human body in standard medicine and is easier to measure. Simplifying the process of rationality evaluation of the physiological curvature of the lumbar vertebra of the human body.
Fourth, the bluetooth data transceiver module of the present invention can upload the measured Cobb angle data to the mobile terminal in real time. By analyzing the data, the daily waist habit information of the user can be obtained, and the severity degree and recurrence probability of the disease can be estimated.
Fifthly, the garment is used as a carrier, and the human body motion state recognition system, the human body posture recognition system, the human lumbar vertebra physiological curvature monitoring system and the signal receiving and transmitting module are all arranged in the intelligent garment, so that a user can conveniently carry the garment in daily work and study, and aesthetic requirements of the user in daily activities are met.
What is not described in detail in this specification is prior art known to those skilled in the art.
Any equivalent transformation based on the technical teaching of the present invention is also within the scope of the present invention.
Claims (4)
1. The intelligent garment is characterized by comprising a garment body, an identification monitoring module and a vibration reminding module, wherein the identification monitoring module and the vibration reminding module are arranged on the garment body, the identification monitoring module is used for collecting the movement state and body posture information of a human body, establishing a human body life posture evaluation model, obtaining a reasonable value range of a Cobb angle under the current movement state and posture through the human body life posture evaluation model, the human body life posture evaluation model is established on the existing medical standard, and the existing medical standard comprises a general standard for referring to the corresponding Cobb angle of the current crowd of different sexes and ages of the current medical students under different movement states and postures; the vibration reminding module is used for reminding the user of taking the waist gesture through vibration;
The recognition monitoring module comprises a depth sensor and a triaxial angular acceleration sensor, wherein the depth sensor comprises a right shoulder joint depth sensor, a left elbow joint depth sensor and a right elbow joint depth sensor which are embedded in the positions of arms and arm joints of the garment, the depth sensors all adopt an ultrasonic ranging principle, the relative distance between the garment joints is obtained by crossing obstacles, a corresponding human body posture prediction model is built by adopting a coordinate method and a distance method and utilizing a convolutional neural network algorithm to carry out data training on the relative distance data, the waist of the garment body is provided with a waistband body conforming to human engineering, the triaxial angular acceleration sensor comprises a first triaxial angular acceleration sensor arranged at the rear collar part of the garment body, a second triaxial angular acceleration sensor embedded at the first lumbar vertebra of the waistband body and a third triaxial angular acceleration sensor embedded at the fifth lumbar vertebra of the waistband body, the first triaxial angular acceleration sensor is used for recognizing the motion state of the human body, and the second triaxial angular acceleration sensor and the third triaxial angular acceleration sensor are respectively used for obtaining the first lumbar vertebra and the fifth lumbar vertebra to carry out physiological monitoring on the human body curvature in real time;
The method for calculating the real-time waist bending Cobb angle of the user is that the vertical earth angle alpha at the first lumbar vertebra and the vertical earth angle beta at the fifth lumbar vertebra are respectively measured by the way of integrating the angular acceleration with time; the difference between the angle beta and the angle alpha can be obtained by subtracting the angle beta of the fifth lumbar vertebra from the angle alpha of the first lumbar vertebra, and the difference is the angle Cobb of the lumbar vertebra measured by the sensor;
the waistband is detachably arranged on the garment body and comprises a waistband body and a composite layer, wherein the composite layer is arranged in the middle of the waistband and comprises a composite material covering layer, a first flexible buffer layer, a medical nylon supporting layer and a second flexible buffer layer which are sequentially arranged from outside to inside;
the first triaxial acceleration sensor is used for identifying the human body motion state, and comprises the steps of training data by utilizing a random forest algorithm to finally obtain a corresponding human body motion state prediction model, and based on the model, the system can identify the real-time motion state of the human body;
the method for establishing the motion state prediction model comprises the following steps:
step 1, a first triaxial acceleration sensor acquires 3-axis original signals tac-XYZ and tBODyGyro-XYZ of an accelerometer and a gyroscope;
Step 2, a median filter and a three-order low-pass Butterworth filter with an angular frequency of 20hz are used for filtering tac-XYZ and tBODIYGyr-XYZ, and the other low-pass Butterworth filter with an angular frequency of 0.3 Hz is used for dividing an acceleration signal tac-XYZ into a body acceleration signal body Acc-XYZ and a gravity acceleration signal tGravityAcc-XYZ;
step 3, deriving the linear acceleration and the angular velocity of the body, and obtaining reflected signals, namely a body acceleration time domain reflected signal tBeodyAccJerk-XYZ returned by an accelerometer and a body angular velocity time domain returned signal tBeodyGyro Jerk-XYZ returned by a gyroscope, calculating the amplitude of the three-dimensional signals by using Euclidean norms, wherein the three-dimensional signals are a body acceleration time domain amplitude signal tBeodyAccMag, a gravity acceleration time domain amplitude signal tGravityAccMag, a body acceleration time domain amplitude reflected signal tBeodyAccJerkMag, a body angular velocity time domain amplitude signal tBeodyGyroMag and a body angular velocity time domain amplitude reflected signal tBeodyGyro JerkMag respectively;
step 4, performing fast Fourier transform on the generated body acceleration time domain signal, body acceleration time domain reflection signal, body angular velocity time domain signal, body acceleration time domain amplitude reflection signal, body angular velocity time domain amplitude signal and body angular velocity time domain amplitude reflection signal, obtaining a body acceleration frequency domain signal tbodyoacc-XYZ, a gravity acceleration frequency domain signal tgovityacc-XYZ, a body acceleration frequency domain reflected signal tbodyojerk-XYZ, a body angular velocity frequency domain signal tbodygoro-XYZ, a body angular velocity frequency domain reflected signal tbodygoro-XYZ, a body acceleration frequency domain amplitude signal tbodyoacmag, a gravity acceleration frequency domain amplitude signal tgovityacmag, a body acceleration frequency domain amplitude reflected signal tbodyorkmag, a body angular velocity frequency domain amplitude signal tbodygromag, a body angular velocity frequency domain amplitude reflected signal tbodygrojerkmag, a body acceleration frequency domain signal fbodyork-XYZ, a body acceleration frequency domain reflected signal fbodygoro-XYZ, a body angular velocity frequency domain signal fbodygoro-XYZ, a body acceleration frequency domain amplitude signal fbodygorMag, a body acceleration frequency domain amplitude reflected signal fbodygorkmag, a body angular velocity frequency domain amplitude signal tbodygorkMag, and a body angular velocity frequency domain amplitude signal tbodygorkMag, wherein the body angular velocity frequency domain amplitude signal tbodymay is represented by a "3 d and a body orientation of the body;
Step 5, obtaining a time domain feature vector and a frequency domain feature vector through the signals in the step 1-3, and recording the motion state information of the human body in different time domains and frequency domains;
and 6, classifying 6 basic motion states of a human body by a hierarchical classification method, namely standing, sitting, prone, walking, going upstairs and going downstairs, wherein in each hierarchy, data comprising time domain feature vectors and frequency domain feature vectors are used as training sets and are input into a random forest classification model for training to obtain a human body motion state prediction model, and based on the model, the system can identify the real-time motion state of the human body.
2. The intelligent garment capable of monitoring and correcting the autonomous posture of lumbar vertebra according to claim 1, wherein the hierarchical classification method specifically comprises the following steps:
step 6.1, classifying static gestures and dynamic gestures in the 1 st level, firstly, uniformly defining a target prediction set corresponding to a sample with a motion state of standing, sitting and prone as a static gesture and datating the static gesture as the same value; the method comprises the steps that a target prediction set corresponding to a sample with a motion state of walking, going upstairs and going downstairs is uniformly specified as a dynamic gesture and is dataized to be another same value, then the dataized target prediction set is subjected to data prediction by using a random forest algorithm to obtain a prediction classification result, and the level 2 is entered after the prediction classification is completed;
Step 6.2, classifying walking and going upstairs and downstairs in the dynamic gestures in the 2 nd level, screening out samples with the target prediction set being the dynamic gesture in the original data set, and uniformly prescribing the target prediction set corresponding to the samples with the dynamic gesture being the walking gesture as the walking gesture and digitizing the walking gesture; uniformly defining target prediction sets corresponding to upstairs and downstairs samples in the dynamic gestures as upstairs and downstairs gestures, and datating the target prediction sets to be the same value, finally, carrying out data training by using a random forest algorithm, wherein the training process is the same as that in the step 6.1, and entering a 3 rd level after the prediction classification is finished;
step 6.3, classifying upstairs and downstairs in upstairs and downstairs postures at the 3 rd level, screening out samples with target prediction sets of upstairs or downstairs in the original data set, inputting the samples of upstairs or downstairs into a random forest algorithm for prediction classification, and entering the 4 th level after the prediction classification is finished;
step 6.4, classifying standing postures and sitting postures in the static postures by the 4 th hierarchy, firstly screening samples with static posture as target prediction sets in the original data set, and then uniformly prescribing the target prediction sets corresponding to the samples with the static postures as standing postures and datating the target prediction sets as the same value; uniformly defining target prediction sets corresponding to standing posture and sitting and lying samples in the static posture as sitting and lying postures and datating the same value; then, carrying out data prediction on the target prediction set after the data processing by using a random forest algorithm to obtain a prediction classification result, and entering a 5 th level after the prediction classification is finished;
And 6.5, classifying sitting postures and prone postures in sitting and lying at the 5 th level, screening out samples of which the target prediction set is the sitting postures or the prone postures in the original data set, inputting the samples of the sitting postures and the prone postures into a random forest algorithm for prediction classification, obtaining a final prediction classification result, finishing the data training process by the prediction classification, and obtaining a human life posture evaluation model.
3. The intelligent garment capable of monitoring and correcting the autonomous posture of lumbar vertebra movement of a human body according to claim 1, wherein the reasonable value range of the Cobb angle in the current movement state and posture is obtained according to the current medical standard after the current movement state and posture are obtained by obtaining the reasonable value range of the Cobb angle in the current movement state and posture through a human body life posture evaluation model.
4. The intelligent garment for monitoring and correcting the autonomous posture of lumbar vertebra according to claim 2, wherein the method for predicting data by using random forest algorithm comprises the following steps:
step 6.11, carrying out replaceable random sampling on a training set containing data of time domain feature vectors and frequency domain feature vectors, randomly taking out m feature vectors each time, and sampling for n times;
Step 6.12, generating 1 decision tree for the feature vector matrix obtained by sampling each time, and generating n decision trees;
step 6.13, based on calculation and judgment modes of the information gain and the information gain rate of the decision tree, an algorithm forms n decision tree prediction models and predicts a target prediction set of the sample;
and 6.14, determining a final prediction classification result in a manner of voting n decision trees.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110886147.5A CN113749644B (en) | 2021-08-03 | 2021-08-03 | Intelligent garment capable of monitoring lumbar vertebra movement of human body and correcting autonomous posture |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110886147.5A CN113749644B (en) | 2021-08-03 | 2021-08-03 | Intelligent garment capable of monitoring lumbar vertebra movement of human body and correcting autonomous posture |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113749644A CN113749644A (en) | 2021-12-07 |
| CN113749644B true CN113749644B (en) | 2023-07-25 |
Family
ID=78788432
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110886147.5A Active CN113749644B (en) | 2021-08-03 | 2021-08-03 | Intelligent garment capable of monitoring lumbar vertebra movement of human body and correcting autonomous posture |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113749644B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114440884A (en) * | 2022-04-11 | 2022-05-06 | 天津果实科技有限公司 | Intelligent analysis method for human body posture for intelligent posture correction equipment |
| CN116369876A (en) * | 2023-04-13 | 2023-07-04 | 中国人民解放军空军特色医学中心 | Pilot distress call device |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR20100093463A (en) * | 2009-02-16 | 2010-08-25 | 전북대학교산학협력단 | Measuring system and method for scoliosis |
| CN104000678A (en) * | 2013-02-24 | 2014-08-27 | 北京银万特科技有限公司 | Intelligent body correcting method and device |
| CN104720821A (en) * | 2015-04-01 | 2015-06-24 | 深圳柔微传感科技有限公司 | Method and smart clothing for achieving real-time posture monitoring |
| KR20180124498A (en) * | 2017-05-12 | 2018-11-21 | 주식회사 본브레테크놀로지 | Scoliosis correction system using wearable IOT device |
| CN109917719A (en) * | 2019-04-04 | 2019-06-21 | 青岛大学 | Pre- myopia prevention dynamic seating position system based on RBF neural |
| KR20200026494A (en) * | 2018-09-03 | 2020-03-11 | 이복동 | Method for Providing Scoliosis Correcting Solution Through Internet |
| CN212089544U (en) * | 2020-02-15 | 2020-12-08 | 杭州普利复健科技有限公司 | Dynamic monitoring device for human spine |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US10624579B2 (en) * | 2018-04-09 | 2020-04-21 | The Hong Kong Polytechnic University | Biofeedback system with body mapping clothing for patients with adolescent idiopathic scoliosis |
-
2021
- 2021-08-03 CN CN202110886147.5A patent/CN113749644B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR20100093463A (en) * | 2009-02-16 | 2010-08-25 | 전북대학교산학협력단 | Measuring system and method for scoliosis |
| CN104000678A (en) * | 2013-02-24 | 2014-08-27 | 北京银万特科技有限公司 | Intelligent body correcting method and device |
| CN104720821A (en) * | 2015-04-01 | 2015-06-24 | 深圳柔微传感科技有限公司 | Method and smart clothing for achieving real-time posture monitoring |
| KR20180124498A (en) * | 2017-05-12 | 2018-11-21 | 주식회사 본브레테크놀로지 | Scoliosis correction system using wearable IOT device |
| KR20200026494A (en) * | 2018-09-03 | 2020-03-11 | 이복동 | Method for Providing Scoliosis Correcting Solution Through Internet |
| CN109917719A (en) * | 2019-04-04 | 2019-06-21 | 青岛大学 | Pre- myopia prevention dynamic seating position system based on RBF neural |
| CN212089544U (en) * | 2020-02-15 | 2020-12-08 | 杭州普利复健科技有限公司 | Dynamic monitoring device for human spine |
Non-Patent Citations (2)
| Title |
|---|
| 穿戴式颈部姿态预防矫正仪设计探讨;韩婉婉;中国高新技术企业(第19期);全文 * |
| 脊椎侧凸及其治疗;俞锦清;浙江中西医结合杂志(第8期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113749644A (en) | 2021-12-07 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Gao et al. | Wearable human motion posture capture and medical health monitoring based on wireless sensor networks | |
| Sun et al. | eButton: a wearable computer for health monitoring and personal assistance | |
| CN113749644B (en) | Intelligent garment capable of monitoring lumbar vertebra movement of human body and correcting autonomous posture | |
| CN104545844B (en) | Multi-parameter sleep monitoring and intelligent diagnosis system based on 4G mobile communication technology and application method of multi-parameter sleep monitoring and intelligent diagnosis system | |
| Olivares et al. | Wagyromag: Wireless sensor network for monitoring and processing human body movement in healthcare applications | |
| Suzuki et al. | Quantitative analysis of motor status in Parkinson’s disease using wearable devices: From methodological considerations to problems in clinical applications | |
| CN109222909B (en) | Wearable intelligent monitoring device and method for monitoring motion, spine bending and joint wear | |
| US20120232430A1 (en) | Universal actigraphic device and method of use therefor | |
| Kang et al. | Real-time elderly activity monitoring system based on a tri-axial accelerometer | |
| CN107530004A (en) | The automatic detection and construction of wearable device based on personal situation, position and/or orientation | |
| US20200085366A1 (en) | Wearable sensor device and analysis platform for objective outcome assessment in spinal diseases | |
| Mazomenos et al. | Detecting elementary arm movements by tracking upper limb joint angles with MARG sensors | |
| CN109171734A (en) | Human body behavioural analysis cloud management system based on Fusion | |
| CN106413529A (en) | Optical pressure sensor | |
| CN105748037A (en) | Body-sensing tank top with biofeedback system for patients with scoliosis | |
| CN109938739A (en) | A kind of cervical vertebra monitoring device | |
| CN107389052A (en) | A kind of ankle pump motion monitoring system and terminal device | |
| CN107194193A (en) | A kind of ankle pump motion monitoring method and device | |
| Jang et al. | Comprehensive monitoring of bad head and shoulder postures by wearable magnetic sensors and deep learning | |
| Nguyen et al. | IMU-based spectrogram approach with deep convolutional neural networks for gait classification | |
| WO2021094775A1 (en) | Method performed by an electronics arrangement for a wearable article | |
| Grecheneva et al. | Estimation of human biomechanics during registration with a wearable device | |
| Ren et al. | Multivariate analysis of joint motion data by Kinect: application to Parkinson’s disease | |
| CN113990439A (en) | Human body posture and physiological information analysis system and working method thereof | |
| CN206301468U (en) | A kind of intelligent garment for children and intelligent children's wearable device |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |