CN105590408B - A kind of tumble detection method for human body and protective device - Google Patents
A kind of tumble detection method for human body and protective device Download PDFInfo
- Publication number
- CN105590408B CN105590408B CN201610083726.5A CN201610083726A CN105590408B CN 105590408 B CN105590408 B CN 105590408B CN 201610083726 A CN201610083726 A CN 201610083726A CN 105590408 B CN105590408 B CN 105590408B
- Authority
- CN
- China
- Prior art keywords
- human body
- axis
- fall
- sampling
- falling
- 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
- 238000001514 detection method Methods 0.000 title claims abstract description 67
- 230000001681 protective effect Effects 0.000 title claims abstract description 5
- 230000001133 acceleration Effects 0.000 claims abstract description 42
- 238000005070 sampling Methods 0.000 claims abstract description 26
- 238000012549 training Methods 0.000 claims abstract description 24
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 6
- 238000012706 support-vector machine Methods 0.000 claims abstract description 4
- 230000006399 behavior Effects 0.000 claims description 28
- 230000000694 effects Effects 0.000 claims description 21
- 230000035484 reaction time Effects 0.000 abstract description 2
- 238000005259 measurement Methods 0.000 abstract 2
- 238000012795 verification Methods 0.000 abstract 2
- 238000000034 method Methods 0.000 description 16
- 238000012360 testing method Methods 0.000 description 9
- 230000008859 change Effects 0.000 description 7
- 230000008569 process Effects 0.000 description 5
- 238000002474 experimental method Methods 0.000 description 4
- 238000004364 calculation method Methods 0.000 description 3
- 238000000605 extraction Methods 0.000 description 3
- 230000035945 sensitivity Effects 0.000 description 3
- 239000013598 vector Substances 0.000 description 3
- 238000005452 bending Methods 0.000 description 2
- 230000005484 gravity Effects 0.000 description 2
- 208000014674 injury Diseases 0.000 description 2
- 239000000203 mixture Substances 0.000 description 2
- 238000005096 rolling process Methods 0.000 description 2
- 230000009466 transformation Effects 0.000 description 2
- 208000012260 Accidental injury Diseases 0.000 description 1
- 206010017076 Fracture Diseases 0.000 description 1
- 206010020100 Hip fracture Diseases 0.000 description 1
- WHXSMMKQMYFTQS-UHFFFAOYSA-N Lithium Chemical compound [Li] WHXSMMKQMYFTQS-UHFFFAOYSA-N 0.000 description 1
- 208000028373 Neck injury Diseases 0.000 description 1
- 206010034719 Personality change Diseases 0.000 description 1
- 208000026137 Soft tissue injury Diseases 0.000 description 1
- 208000027418 Wounds and injury Diseases 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 208000029028 brain injury Diseases 0.000 description 1
- 238000013145 classification model Methods 0.000 description 1
- 230000000052 comparative effect Effects 0.000 description 1
- 239000002131 composite material Substances 0.000 description 1
- 238000004590 computer program Methods 0.000 description 1
- 230000006378 damage Effects 0.000 description 1
- 238000013480 data collection Methods 0.000 description 1
- 230000003247 decreasing effect Effects 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 239000002360 explosive Substances 0.000 description 1
- 229910052744 lithium Inorganic materials 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 238000003786 synthesis reaction Methods 0.000 description 1
- 238000000844 transformation Methods 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
- 210000000707 wrist Anatomy 0.000 description 1
Classifications
-
- G—PHYSICS
- G08—SIGNALLING
- G08B—SIGNALLING OR CALLING SYSTEMS; ORDER TELEGRAPHS; ALARM SYSTEMS
- G08B21/00—Alarms responsive to a single specified undesired or abnormal condition and not otherwise provided for
- G08B21/02—Alarms for ensuring the safety of persons
- G08B21/04—Alarms for ensuring the safety of persons responsive to non-activity, e.g. of elderly persons
- G08B21/0407—Alarms for ensuring the safety of persons responsive to non-activity, e.g. of elderly persons based on behaviour analysis
- G08B21/043—Alarms for ensuring the safety of persons responsive to non-activity, e.g. of elderly persons based on behaviour analysis detecting an emergency event, e.g. a fall
-
- G—PHYSICS
- G08—SIGNALLING
- G08B—SIGNALLING OR CALLING SYSTEMS; ORDER TELEGRAPHS; ALARM SYSTEMS
- G08B21/00—Alarms responsive to a single specified undesired or abnormal condition and not otherwise provided for
- G08B21/02—Alarms for ensuring the safety of persons
- G08B21/04—Alarms for ensuring the safety of persons responsive to non-activity, e.g. of elderly persons
- G08B21/0438—Sensor means for detecting
- G08B21/0446—Sensor means for detecting worn on the body to detect changes of posture, e.g. a fall, inclination, acceleration, gait
Landscapes
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Gerontology & Geriatric Medicine (AREA)
- Business, Economics & Management (AREA)
- Emergency Management (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Psychiatry (AREA)
- Psychology (AREA)
- Social Psychology (AREA)
- Measurement Of The Respiration, Hearing Ability, Form, And Blood Characteristics Of Living Organisms (AREA)
Abstract
The invention discloses a kind of tumble detection method for human body and protective device; obtain three axis resultant accelerations of each sampling time point, roll angle, pitch angle; sliding time window is set; obtain the average and standard deviation of three axis resultant accelerations of each sampled point in each time window; the sum of absolute value of variable quantity of roll angle and pitch angle when last sample point human body of sliding time window is stood relatively is obtained, characteristic value is used as using three parameters；Grader is built using algorithm of support vector machine；Tumble sample and daily routines behavior sample composing training collection are obtained, grader is trained, the grader after being trained；Fall detection is carried out according to the actual sensing data of human body of acquisition using the grader after training.The verification and measurement ratio of tumble behavior of the present invention is 99.2%, and the verification and measurement ratio of daily routines behavior is 96%, and the average lead time reaches 273ms, for the real-time early warning of tumble and the reaction time for starting offer of protective device.
Description
Technical Field
The invention relates to a detection method for human body protection and a protection device, in particular to a detection method for human body road fall and a device for protecting a human body when the human body falls.
Background
Falls are a common accident for the elderly, who often have injuries caused by falls, such as hip fractures, neck injuries, brain injuries, and various soft tissue injuries and other fractures.
The detection of human body falling events is a precondition for protection. In the prior art, there are mainly 3 types of human body fall detection methods: video image based detection, audio or radio based detection of ambient signals and wearable device based detection. The video image detection has the advantages that the human body does not need to carry any equipment, and the video image detection has the disadvantages that the video image is greatly influenced by light, environment and the like, the detection range is limited and the personal privacy is involved. The method based on ambient signal detection is greatly influenced by ambient environment, cannot obtain high precision, and generally can only be used as an auxiliary detection method. The method based on the wearable device detection is not influenced by the surrounding environment due to low cost and large detection range, and is the fall detection method which is most researched and practically applied at present.
For example, chinese patent application CN102117533A discloses a human fall detection deviceThe detection protection and alarm device comprises a falling detection part, a falling detection information transmission part, a falling position detection part and a protection air bag. The wearable device is used to try to start the protective airbag by acquiring a fall detection signal, but the detection is performed by a method of setting a threshold value, as in most current fall detection methods. It adopts an acceleration sensor and a gyroscope, defines a signal vector modulus SVM and a motion angular velocity,(namely three-axis resultant acceleration), an SVM threshold value is set to be 1.8-2.2 g, a motion angular velocity threshold value is 0.5rad/s-0.55rad/s, and when the SVM and the human motion angular velocity exceed the set threshold value, the human body is judged to fall. The threshold detection method belongs to post detection, so that the fall protection device is difficult to play a role.
The human body state is judged by a method of training a classifier, so that the falling trend can be detected earlier than the common threshold value method, and the prior detection is realized. However, how to select the feature values used by the classifier, and what classifier is used, are key points for actually realizing the prior detection.
Therefore, how to detect the human body falling in advance is an urgent problem to be solved for realizing human body falling protection at present.
Disclosure of Invention
The invention aims to provide a human body falling detection method, which realizes the prior detection of human body falling so as to provide enough reaction time for a protection device; another object of the present invention is to provide a human fall protection device using such a detection method.
In order to achieve the purpose of the invention, the technical scheme adopted by the invention is as follows: a human fall detection method, comprising:
1) obtaining a characteristic value:
sampling by using a three-axis accelerometer and a three-axis gyroscope worn by a human body to obtain sampling data of each sampling time point;
②, respectively calculating and obtaining the three-axis resultant acceleration of each sampling time point according to the sampling data obtained in the step ①Angle of rollAnd a pitch angle，
Wherein,，a x 、a y 、a z three axes of acceleration, x and y axes in two orthogonal directions in the horizontal plane, and roll angleIs the attitude angle, pitch angle about the x-axisIs the attitude angle about the y-axis;
setting the length of the sliding time window and the superposition rate of the adjacent sliding time windows to obtain the average value of the three-axis combined acceleration of each sampling point in each sliding time windowAnd standard deviation ofObtaining the sum of the absolute values of the variation of the rolling angle and the pitching angle of the human body relative to the vertical state at the last sampling point of the sliding time window, and taking the three values as the basisTaking the parameters as characteristic values;
2) establishing and training a classifier:
constructing a classifier by adopting a support vector machine algorithm, and taking the three parameters obtained in the step 1) as input characteristic values of the classifier;
respectively carrying out daily activity behaviors and different falling behaviors by trained personnel, obtaining falling samples and daily activity behavior samples to form a training set, and training the classifier to obtain the trained classifier;
3) and carrying out falling detection by adopting the trained classifier according to the obtained actual sensing data of the human body.
in ① of above technical solution, in ① of step 1), ① of sampling frequency is usually related to ① of performance of ① of adopted sensor chip, and too low sampling frequency will result in insufficient number of sampling points in ① of sliding time window, thereby affecting ① of judgment effect.
The preferred sampling frequency is 100 Hz.
In the technical scheme, the length of the sliding time window is 100-300 ms, and the overlapping rate of the window is 40-60%.
The length of the sliding time window is 100ms, and the overlapping rate of the window is 50%.
The invention also provides a human body falling protection device which comprises a falling detection device, a protection air bag and a driving device, wherein the falling detection device mainly comprises a three-axis accelerometer, a three-axis gyroscope and a controller, and the trained classifier is arranged in the controller.
According to the preferable technical scheme, the falling detection device is arranged at the waist of a human body.
In the above technical solution, the driving device includes a compressed gas cylinder communicated with the protection airbag through a control valve, and the control valve is controlled by the controller to open and close.
The protection airbag is composed of a plurality of airbags respectively worn on the fragile parts of people.
Due to the application of the technical scheme, compared with the prior art, the invention has the following advantages:
1. the invention selects three parameters respectively related to the acceleration and the angular velocity, and adopts a support vector machine algorithm to train a classifier, thereby realizing the prior detection of human body falling, and the experimental result shows that the detection rate of the falling behavior is 99.2 percent, the detection rate of the daily activity behavior is 96 percent, the detection rate of the violent daily activity behavior is higher, the average lead time reaches 273ms, and the invention provides the response time for the real-time early warning of falling and the starting of the protection device.
2. The invention adopts the inflation mode of the compressed gas cylinder to inflate the air bag, avoids the accidental injury caused by the instant impact force caused by the explosive starting in the prior art, and the air bag can be used as wearing equipment to be arranged at important positions of a human body in clothes, such as hip, head and neck and other key positions, thereby playing a good role in protecting the human body.
Drawings
FIG. 1 is a flow chart of a method of an embodiment of the present invention;
FIG. 2 is a graph of the resultant acceleration of a forward fall in an embodiment;
FIG. 3 is a roll angle variation graph of a forward fall in an embodiment;
fig. 4 is an acceleration vector and a change curve of a walking process in the second embodiment.
Detailed Description
The invention is further described with reference to the following figures and examples:
example (b): referring to fig. 1, the present embodiment provides a human fall detection method, including: and acquiring training data, performing feature extraction to obtain a training sample, and training the classification model to obtain the trained classifier. And in order to detect the effect, test data is obtained, feature extraction is carried out, and a trained classifier is adopted for classification decision.
When feature extraction is performed, factors specifically considered for selecting feature values are expressed as follows:
the sampling frequency in this example is 100 Hz.
(1) Selection of acceleration characteristics:
in the process of falling down, the acceleration can obviously change, and the three-axis combined acceleration is taken in consideration of different falling directions of front, back, left and right:
；
fig. 2 is a resultant acceleration curve of a forward fall, and it can be seen that, in the process of a fall, the resultant acceleration is first decreased and then increased, and the moment when the acceleration is maximum is the moment when the ground is collided, and the fall detection in advance is to detect the fall before the moment. The time from the detection of a fall to the impact on the ground is called lead time. Accordingly, when training and testing the SVM model, the collection of the fall sample should be performed before the moment and a certain time is reserved.
Taking the magnitude and the variation trend of the resultant acceleration into consideration, taking a sliding time window with the length of n (n sampling points), and extracting the following acceleration characteristics:
(1) mean of resultant accelerations:
and i is the serial number of the sampling point in a certain sliding time window.
(2) Standard deviation of resultant acceleration:
in this embodiment, n is 10, that is, the length of the sliding time window is 100ms, and the overlap ratio of the windows is 50%, that is, the last 50% of data of the current window is used as the first 50% of data of the next window.
(2) Selection of angular features:
in the falling process of the human body, the attitude angle of the human body also changes obviously, a three-dimensional coordinate system of the human body is established by taking the east direction as an x axis, the north direction as a y axis and the upward direction as a z axis, the three-dimensional coordinate system is also called as a northeast sky coordinate system, and the attitude angles around the x axis, the y axis and the z axis are respectively called as a rolling angle gamma, a pitch angle theta and a yaw angle psi. When the human body falls back and forth, gamma changes, and when the human body falls left and right, theta changes. When the human body moves violently, the sensor cannot distinguish the gravity acceleration from the self acceleration, the attitude angle cannot be directly obtained through the acceleration, the attitude calculation is required, and a commonly used attitude calculation method is a quaternion method.
Quaternion is a number made up of four elements:
the quaternion method introduces quaternion in algebra as a tool to make up for the deficiency of describing the angular motion of a rigid body by Euler angles. The basis of attitude calculation is coordinate transformation, any attitude change can be considered as the angular synthesis through three times of axial rotation in the moment, the rotation sequence cannot be changed, and the spatial attitude of the human body can be regarded as a composite result after basic rotation is carried out around the z axis, the y axis and the x axis in sequence. After a series of mathematical transformations, the attitude angle expressed by quaternion is:
。
therefore, if the quaternion Q is determined, the roll angle γ and pitch angle θ can be calculated. Mathematically derived, the quaternion differential equation can be expressed as:
wherein:，、、expressing the three-axis angular velocity, solving the differential equation by adopting a Picard algorithm to obtain:
wherein: i is a unit matrix of the image data,，，、、is the x, y and z axes are onThe angular increment within the time interval is sampled.
Fig. 3 shows the curve of the roll angle for a forward fall, and the final angle is the angle at the moment of impact with the ground, and it can be seen that the change of the angle is close to 90 °, and likewise, the change of the angle is close to 90 ° when the person falls backward, leftward and rightward. The sum of the absolute values of the amounts of change in roll angle and pitch angle with respect to standing is taken as an angular feature, taking into account the different directions of falls.
Thus, a total of 3 features, the mean value and standard deviation of the acceleration within the sliding time window, and the sum of the absolute values of the amounts of change in the roll angle and pitch angle with respect to standing, are extracted.
To verify the method of this example, a wearable device was prepared. Wearable equipment comprises MEMS sensor module, bluetooth module and lithium cell. The MEMS sensor module includes an inertial sensor MPU6050, which MPU6050 itself integrates a three-axis accelerometer and a three-axis gyroscope. The sensor module respectively collects triaxial acceleration and triaxial angular velocity at the frequency of 100Hz and transmits the triaxial acceleration and the triaxial angular velocity to the PC upper computer through the Bluetooth module, and the PC upper computer adopts Visual C + +6.0 to write data collection and data processing programs.
Since the daily activities of the head, the wrist, the upper arm and other parts of the human body are frequent, the sensor is worn at the parts, which easily causes false alarm, the subjective feeling of a wearer is affected when the sensor is placed on the chest, the ergonomic data shows that the gravity center position of the human body is about 56% of the height of the human body in the upright state, so the sensor is worn at the waist of the human body properly, and the interference caused by the daily activities of the human body is small.
The device is adopted to collect the characteristics.
And respectively collecting a falling sample and a daily activity behavior sample to carry out SVM model training and testing. This embodiment defines 4 fall patterns: fall forward, fall backward, fall left and fall right. Define 10 activities of daily living: standing, sitting, standing from sitting, squatting, standing from squatting, bending over, lying, sitting from lying, walking and jogging. The number of the experiment persons is 5, the age is 23 + -1.3, the height is 171.4 + -2.2 cm, the weight is 61.2 + -4.6 kg, and the falling experiment is completed on a sponge cushion with the thickness of 4m multiplied by 1m multiplied by 0.2 m. The obtained fall samples and daily activity behavior samples are divided into a training sample set and a testing sample set, and the composition of the sample sets is shown in table 1.
TABLE 1 composition of sample set
| Sample set | Number of samples for tumble | Number of samples of daily activities |
| Training set | 50 | 50 |
| Test set | 250 | 250 |
The training sample set is used for training the SVM model, and the testing sample set is used for evaluating the accuracy of classification. The SVM algorithm is realized by using an LIBSVM development kit. The accuracy of fall detection can be measured by the following 2 indicators:
wherein: TP represents the number of samples in which a fall occurred and was detected, FN represents the number of samples in which a fall occurred but was missed, TN represents the number of samples in which a daily behavior occurred and was detected, and FP represents the number of samples in which a daily behavior was mistaken for a fall.
Reserving different time to collect a falling sample, comparing the conditions of only adopting acceleration characteristics, only adopting angle characteristics and combining the acceleration characteristics and the angle characteristics, and performing model training and testing, wherein the experimental result is as follows:
TABLE 2 use acceleration characteristics only
| Index (I) | 100ms | 150ms | 200ms | 250ms | 300ms |
| Sen/% | 98.8 | 98 | 98 | 99.2 | 99.2 |
| Spe/% | 65.2 | 69.2 | 71.6 | 64.8 | 42 |
Note: sen = Sensitivity, Spe = Sensitivity
As can be seen from table 2, the accuracy of the model obtained by training the fall sample collected for 200ms is the best when only the acceleration feature is used.
TABLE 3 angular characteristics taken only
| Index (I) | 100ms | 150ms | 200ms | 250ms | 300ms |
| Sen/% | 99.2 | 99.2 | 99.2 | 99.2 | 99.2 |
| Spe/% | 92 | 91.6 | 88.8 | 74 | 75.6 |
As can be seen from table 3, the accuracy of the model obtained by training the fall sample collected for 100ms is the best when only the angle features are used.
TABLE 4 combines acceleration and angular characteristics
| Index (I) | 100ms | 150ms | 200ms | 250ms | 300ms |
| Sen/% | 99.2 | 99.2 | 99.2 | 99.2 | 99.2 |
| Spe/% | 95.6 | 92.8 | 96 | 86 | 76 |
As can be seen from table 4, when the acceleration characteristic and the angle characteristic are combined, the model accuracy obtained by training the fall sample collected for 200ms is the best. The optimal test results for these three cases are compared as shown in the following table:
TABLE 5 comparative results
| Index (I) | Acceleration only | Angle only | Acceleration and angle |
| Sen/% | 98 | 99.2 | 99.2 |
| Spe/% | 71.6 | 92 | 96 |
As can be seen from table 5, the combined acceleration and angular characteristics result better than using a single acceleration or angular characteristic. Analysis data shows that when only the acceleration characteristic is adopted, daily behaviors with severe acceleration changes such as rapid squatting and jogging are easily mistaken for falling behaviors, and when only the angle characteristic is adopted, daily behaviors with large angle changes such as bending and lying are easily mistaken for falling behaviors.
The optimal SVM model is imported into a PC upper computer program, a 100ms sliding time window and a 50% window overlapping rate are set, the lead time of 250 groups of falling test samples is counted, and the result is shown in Table 6.
TABLE 6 lead time statistics
| Lowest level of | Highest point of the design | Average |
| 130ms | 570ms | 273ms |
It can be seen that, in this embodiment, by using the method of the present invention, the detection rate of the falling behavior is 99.2%, the detection rate of the daily activity behavior is 96%, for a more strenuous daily activity behavior, the detection rate is also higher, the average lead time reaches 273ms, and time is provided for the real-time early warning of falling and the start of the protection device.
Example two: the training experiment was performed by varying the length of the sliding window using the method described in example one.
Referring to FIG. 4, acceleration vectors and variation curves of a walking process are defined, a window length w and a window overlap length o are defined, and a time series data is obtainedThe first window is represented asAcceleration and angular velocity features are extracted for the data within this window. Since the overlap length of the window is o, the data of the next window is represented asAnd the data in the window is processed continuously, and so on, so that a plurality of groups of daily behavior samples can be collected in one daily behavior. In the experiment, the completion time of each daily activity is 5 seconds, the length w of a sliding window for collecting daily activity samples is also 100ms, the window overlapping rate is 50 percent, namely the last 50 percent of data of the current window is used as the first 50 percent of data of the next window, samples with severe data change, namely samples which are easy to be mistakenly judged to fall, are selected from the data, and 300 groups of daily activity samples are obtained.
Sampling is carried out by adopting different lengths w of the sliding window, in the actual sample of the falling behavior, the reserved time from the sliding window for sampling to the actual falling is t, and the result is shown in the following table:
| w/ms | t/ms | Sensitivity/% | Specificity/% | lead time/ms |
| 100 | 200 | 99.6 | 96.8 | 283 |
| 150 | 150 | 100 | 96 | 269 |
| 200 | 200 | 100 | 78 | 296 |
| 250 | 150 | 99.6 | 100 | 271 |
| 300 | 150 | 99.6 | 99.6 | 266 |
As can be seen from the above table, when w is 250ms and the reserved time t is 150ms, the accuracy of the fall detection model obtained by sample training is the best, the fall detection model is used as the optimal fall detection model, the detection rate of the fall behavior is 99.6%, only 1 group of fall behaviors are detected before the fall collides with the ground, the detection rate of the daily activity behavior is 100%, no misjudgment occurs, and the detection rate is also high for the relatively violent daily activity behaviors such as jogging. The mean lead time for fall detection was 271ms, and the lead time distribution of fall behavior detected in 249 sets is shown in the table below.
Front time distribution
| Preamble time distribution interval | Number of sample groups |
| <100ms | 0 |
| 100~200ms | 15 |
| 200~300ms | 185 |
| 300~400ms | 46 |
| >400ms | 3 |
Therefore, the fall detection device can effectively detect falls, has enough lead time, and can be used for controlling the human body fall protection device.
Claims (9)
1. A human fall detection method, comprising:
1) obtaining a characteristic value:
sampling by using a three-axis accelerometer and a three-axis gyroscope worn by a human body to obtain sampling data of each sampling time point;
②, respectively calculating and obtaining the three-axis resultant acceleration of each sampling time point according to the sampling data obtained in the step ①Angle of rollAnd a pitch angle，
Wherein,，a x 、a y 、a z three axes of acceleration, x and y axes in two orthogonal directions in the horizontal plane, and roll angleIs the attitude angle, pitch angle about the x-axisIs the attitude angle about the y-axis;
setting the length of the sliding time window and the superposition rate of the adjacent sliding time windows to obtain the average value of the three-axis combined acceleration of each sampling point in each sliding time windowAnd standard deviation ofObtaining the sum of absolute values of the variation of the roll angle and the pitch angle of the human body relative to the upright state at the last sampling point of the sliding time window, and taking the three parameters as characteristic values;
2) establishing and training a classifier:
constructing a classifier by adopting a support vector machine algorithm, and taking the three parameters obtained in the step 1) as input characteristic values of the classifier;
respectively carrying out daily activity behaviors and different falling behaviors by trained personnel, obtaining falling samples and daily activity behavior samples to form a training set, and training the classifier to obtain the trained classifier;
3) and carrying out falling detection by adopting the trained classifier according to the obtained actual sensing data of the human body.
2. the human fall detection method according to claim 1, wherein the sampling frequency in step 1) of ① is not less than 50 Hz.
3. A human fall detection method according to claim 2, wherein: the sampling frequency was 100 Hz.
4. A human fall detection method according to claim 1, wherein: the length of the sliding time window is 100-300 ms, and the overlapping rate of the window is 40% -60%.
5. A human fall detection method according to claim 1, wherein: the length of the sliding time window is 100ms, and the overlapping rate of the window is 50%.
6. A human body falling protection device comprises a falling detection device, a protection air bag and a driving device, and is characterized in that: the fall detection device mainly comprises a three-axis accelerometer, a three-axis gyroscope and a controller, wherein the trained classifier in claim 1 is arranged in the controller.
7. A personal fall protection device according to claim 6, wherein: the fall detection device is arranged at the waist of a human body.
8. A personal fall protection device according to claim 6, wherein: the driving device comprises a compressed gas cylinder communicated with the protective gas bag through a control valve, and the control valve is controlled by the controller to open and close.
9. A personal fall protection device according to claim 6, wherein: the protection airbag is composed of a plurality of airbags respectively worn on the fragile parts of the human body.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610083726.5A CN105590408B (en) | 2016-02-06 | 2016-02-06 | A kind of tumble detection method for human body and protective device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610083726.5A CN105590408B (en) | 2016-02-06 | 2016-02-06 | A kind of tumble detection method for human body and protective device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105590408A CN105590408A (en) | 2016-05-18 |
| CN105590408B true CN105590408B (en) | 2018-05-04 |
Family
ID=55929958
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610083726.5A Active CN105590408B (en) | 2016-02-06 | 2016-02-06 | A kind of tumble detection method for human body and protective device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105590408B (en) |
Families Citing this family (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106227999B (en) * | 2016-07-16 | 2018-10-09 | 北京工业大学 | A kind of high-adaptability tumble behavioral value method |
| CN106539587A (en) * | 2016-12-08 | 2017-03-29 | 浙江大学 | A kind of fall risk assessment and monitoring system and appraisal procedure based on sensor of doing more physical exercises |
| CN106683340A (en) * | 2016-12-15 | 2017-05-17 | 歌尔股份有限公司 | User behavior monitoring method and wearable equipment |
| CN106910314B (en) * | 2017-02-03 | 2019-01-25 | 同济大学 | A kind of personalized fall detection method based on the bodily form |
| CN107495972A (en) * | 2017-08-14 | 2017-12-22 | 哈尔滨工业大学（威海） | One kind falls down detection algorithm |
| CN107933476B (en) * | 2017-11-14 | 2019-08-13 | 苏州数言信息技术有限公司 | The method and system of the sensing device of the anti-forgetting of general passenger inside the vehicle |
| CN108158591A (en) * | 2017-12-26 | 2018-06-15 | 深圳市康莱米电子股份有限公司 | Detect the method and device fallen down |
| CN109993312B (en) * | 2018-01-02 | 2021-04-27 | 中国移动通信有限公司研究院 | Equipment, information processing method thereof and computer storage medium |
| CN108549900A (en) * | 2018-03-07 | 2018-09-18 | 浙江大学 | Tumble detection method for human body based on mobile device wearing position |
| CN109064710A (en) * | 2018-08-14 | 2018-12-21 | 东南大学 | With the security protection terminal and system for falling down identification function |
| CN109833048B (en) * | 2019-01-23 | 2021-11-05 | 中国民航大学 | Alertness measuring method based on psychomotor ability |
| CN109979162A (en) * | 2019-03-27 | 2019-07-05 | 苏州威斯德医疗科技有限公司 | A kind of anti-tumble alarm method and alarm system |
| CN110946585A (en) * | 2019-11-21 | 2020-04-03 | 上海理工大学 | Fall detection system and method based on data fusion and BP neural network |
| CN112472075A (en) * | 2020-12-21 | 2021-03-12 | 福州数据技术研究院有限公司 | Fall detection method and system and storage device |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN202650197U (en) * | 2012-07-09 | 2013-01-02 | 台州学院 | Fall prevention apparatus |
| CN104224182A (en) * | 2014-03-31 | 2014-12-24 | 桂林电子科技大学 | Method and device for monitoring human tumbling |
| CN104586398A (en) * | 2013-10-30 | 2015-05-06 | 广州华久信息科技有限公司 | Old man falling detecting method and system based on multi-sensor fusion |
| CN104622454A (en) * | 2015-01-23 | 2015-05-20 | 深圳市卡麦睿科技有限公司 | Multi-functional bracelet type human body intelligent monitor system |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US9402568B2 (en) * | 2011-08-29 | 2016-08-02 | Verizon Telematics Inc. | Method and system for detecting a fall based on comparing data to criteria derived from multiple fall data sets |
-
2016
- 2016-02-06 CN CN201610083726.5A patent/CN105590408B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN202650197U (en) * | 2012-07-09 | 2013-01-02 | 台州学院 | Fall prevention apparatus |
| CN104586398A (en) * | 2013-10-30 | 2015-05-06 | 广州华久信息科技有限公司 | Old man falling detecting method and system based on multi-sensor fusion |
| CN104224182A (en) * | 2014-03-31 | 2014-12-24 | 桂林电子科技大学 | Method and device for monitoring human tumbling |
| CN104622454A (en) * | 2015-01-23 | 2015-05-20 | 深圳市卡麦睿科技有限公司 | Multi-functional bracelet type human body intelligent monitor system |
Non-Patent Citations (2)
| Title |
|---|
| 基于Android平台的跌倒检测软件设计;刘磊;《万方数据》;20151229;第8-17页 * |
| 基于多传感器的人体运动模式识别研究;李路;《万方数据》;20131030;第31-43页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105590408A (en) | 2016-05-18 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105590408B (en) | A kind of tumble detection method for human body and protective device | |
| AU2020273327B2 (en) | Systems and methods of swimming analysis | |
| CN104056441B (en) | Information processing system, information processing method, and storage medium | |
| Shi et al. | Mobile human airbag system for fall protection using MEMS sensors and embedded SVM classifier | |
| JP5437549B2 (en) | Activity pattern monitoring method and apparatus | |
| CN106981174A (en) | A kind of Falls Among Old People detection method based on smart mobile phone | |
| CN111700624B (en) | Pattern recognition method and system for detecting motion gesture by intelligent bracelet | |
| CN107169512B (en) | HMM-SVM falling model construction method and falling detection method based on model | |
| JP2009525107A (en) | Gait analysis | |
| JP2013537436A (en) | Integrated portable device and method implementing accelerometer for analyzing stride biomechanical parameters | |
| KR101872907B1 (en) | Motion analysis appratus and method using dual smart band | |
| US10894186B2 (en) | Real time golf swing training aid | |
| JP6241488B2 (en) | Exercise support device, exercise support method, and exercise support program | |
| CN105944354A (en) | A basketball shooting monitoring auxiliary device | |
| KR101651429B1 (en) | Fitness monitoring system | |
| CN107662613B (en) | A kind of extreme driving behavior recognition methods and system based on mobile intelligent perception | |
| CN106256396A (en) | Motion assisting system and motion support method | |
| CN108338790B (en) | Gait analysis and fall assessment system | |
| Mack et al. | Amateur boxer biomechanics and punch force | |
| CN115462782B (en) | Human body falling dynamic monitoring method and system based on multidimensional characteristic parameters | |
| US20200193865A1 (en) | Real time sports motion training aid | |
| CN113663312A (en) | Micro-inertia-based non-apparatus body-building action quality evaluation method | |
| CN114469074A (en) | Fall early warning method, system, equipment and computer storage medium | |
| CN113975775B (en) | Wearable inertial body feeling ping-pong exercise training system and working method thereof | |
| US20220193522A1 (en) | Exercise analysis system using sensor worn on user's head |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |